WO2022007859A1 - 激光雷达参数调整方法、装置及激光雷达 - Google Patents
激光雷达参数调整方法、装置及激光雷达 Download PDFInfo
- Publication number
- WO2022007859A1 WO2022007859A1 PCT/CN2021/105048 CN2021105048W WO2022007859A1 WO 2022007859 A1 WO2022007859 A1 WO 2022007859A1 CN 2021105048 W CN2021105048 W CN 2021105048W WO 2022007859 A1 WO2022007859 A1 WO 2022007859A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- lidar
- parameter adjustment
- scene
- adjustment strategy
- area
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000013528 artificial neural network Methods 0.000 claims description 60
- 238000001514 detection method Methods 0.000 claims description 56
- 230000003287 optical effect Effects 0.000 claims description 30
- 238000004891 communication Methods 0.000 claims description 28
- 230000005540 biological transmission Effects 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 7
- 230000007613 environmental effect Effects 0.000 claims description 6
- 238000013519 translation Methods 0.000 description 15
- 238000010586 diagram Methods 0.000 description 14
- 238000012549 training Methods 0.000 description 14
- 230000003247 decreasing effect Effects 0.000 description 6
- 238000004590 computer program Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 241000283070 Equus zebra Species 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000000429 assembly Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/497—Means for monitoring or calibrating
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/89—Lidar systems specially adapted for specific applications for mapping or imaging
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
- G01S17/933—Lidar systems specially adapted for specific applications for anti-collision purposes of aircraft or spacecraft
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/4802—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/4808—Evaluating distance, position or velocity data
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
- G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
Definitions
- Embodiments of the present invention relate to the technical field of radar, and in particular, to a method and device for adjusting parameters of a laser radar, and a laser radar.
- lidar has been widely used in the fields of automatic driving and assisted driving.
- One objective of the embodiments of the present invention is to provide a method, device and a laser radar for adjusting the parameters of the laser radar, which can automatically adjust the working parameters of the laser radar according to different scenarios.
- a method for adjusting parameters of a lidar including: acquiring three-dimensional environment information around a lidar; and identifying a scene type and a drivable area where the lidar is located based on the three-dimensional environment information ; Determine a parameter adjustment strategy of the lidar according to the scene type and the drivable area, and adjust the current working parameters of the lidar based on the parameter adjustment strategy.
- the 3D environment information includes 3D point cloud data; the identifying the scene type and the drivable area where the lidar is located based on the 3D environment information specifically includes:
- the three-dimensional point cloud data is processed to generate a multi-channel point cloud feature map; high-dimensional feature information is extracted from the multi-channel point cloud feature map; according to the high-dimensional feature information, the scene type is determined; according to the high-dimensional feature information to determine the drivable area.
- the extracting high-dimensional feature information from the multi-channel point cloud feature map specifically includes: inputting the multi-channel point cloud feature map into a first neural network, and obtaining the first the high-dimensional feature information output by the neural network;
- the determining the scene type according to the high-dimensional feature information specifically includes: inputting the high-dimensional feature information into a second neural network, and acquiring a scene type output value output by the second neural network; according to a preset The correspondence between the scene type output value and the scene type label, to determine the scene type corresponding to the scene type output value;
- the determining of the drivable area according to the high-dimensional feature information specifically includes: inputting the high-dimensional feature information into a third neural network, and obtaining an output map of the drivable area output by the third neural network; The corresponding relationship between the drivable area output map and the drivable area label is preset, and the drivable area corresponding to the drivable area output map is determined.
- the parameter adjustment strategy of the lidar is determined according to the scene type and the drivable area, and the current working parameters of the lidar are adjusted based on the parameter adjustment strategy, Specifically, it includes: according to the scene type, determining the horizontal field of view angle of the lidar, the vertical field of view of the lidar, the optical axis direction of the emitted beam of the lidar, the scanning density of the lidar, A scene parameter adjustment strategy for one or more of the scanning frequency of the lidar and the pulse transmission power of the lidar; according to the drivable area, the scene parameter adjustment strategy is adjusted, and the adjusted The scene parameter adjustment strategy is determined as the parameter adjustment strategy of the lidar, and the current working parameters of the lidar are adjusted based on the parameter adjustment strategy.
- the scene parameter adjustment strategy is adjusted according to the drivable area, and the adjusted scene parameter adjustment strategy is determined as the lidar parameter adjustment strategy, and based on The parameter adjustment strategy adjusts the current working parameters of the lidar, which specifically includes: obtaining, according to the scene type, a standard driving area under the scene type and a horizontal field of view of the lidar corresponding to the standard driving area and the driving parameter adjustment strategy of the vertical field of view of the lidar; the standard driving area and the drivable area are matched, and the scene is adjusted according to the matching result and the driving parameter adjustment strategy A parameter adjustment strategy; the current working parameters of the lidar are adjusted based on the adjusted scene parameter adjustment strategy.
- the standard driving area and the drivable area are matched, and the scene parameter adjustment strategy is adjusted according to the matching result and the driving parameter adjustment strategy, which specifically includes : analyze the topological structure of the drivable area and the topological structure of the standard driving area; calculate the topological correlation between the topological structure of the drivable area and the topological structure of the standard driving area; according to the topological correlation, Determine the scene difference coefficient of the drivable area relative to the standard travel area; adjust the scene parameter adjustment strategy according to the scene difference coefficient and the driving parameter adjustment strategy.
- the current working parameters of the lidar include: the vertical field of view angle of the emitted beam of the lidar, the horizontal field of view of the emitted beam of the lidar, the One or more of the optical axis direction of the emitted light beam, the scanning frequency of the lidar, and the pulse emission frequency of the lidar.
- a lidar parameter adjustment device including: an environment information acquisition module for acquiring three-dimensional environment information around the lidar; an identification module for identifying, based on the three-dimensional environment information, The scene type and the drivable area where the lidar is located; the adjustment module is used to determine the parameter adjustment strategy of the lidar according to the scene type and the drivable area, and adjust all the parameters based on the parameter adjustment strategy. Describe the current operating parameters of the lidar.
- a laser radar including: a transmitting device, a receiving device, a processor, a memory, a communication interface, and a communication bus, the processor, the memory, and the communication interface passing through The communication bus completes mutual communication; the transmitting device is used for transmitting outgoing laser light to the detection area; the receiving device is used for receiving the echo laser reflected by the object in the detection area; the memory is used for storing at least An executable instruction that causes the processor to perform the steps of the method for adjusting the parameters of the lidar as described above.
- a method for adjusting lidar parameters is provided, which is applied to a lidar system including multiple lidars, including:
- the scene type determine the horizontal field of view of at least one of the lidars in the lidar system, the vertical angle of view of the lidar, the optical axis direction of the emitted beam of the lidar, the laser beam
- Adjust the scene parameter adjustment strategy of at least one of the lidars according to the drivable area determine the adjusted scene parameter adjustment strategy as the parameter adjustment strategy of at least one of the lidars, and adjust the parameters based on the parameters
- the strategy adjusts the current operating parameters of at least one of the lidars.
- the acquiring three-dimensional environment information around the lidar system includes:
- the lidar is a lidar that detects a mid-to-long-range detection field of view;
- the three-dimensional environment information around the lidar system is obtained through a plurality of lidars in the lidar system, wherein the plurality of lidars correspond to different detection fields of view, and the plurality of lidars are integrated to obtain the information respectively. to obtain the complete three-dimensional environment information of the lidar system.
- the method further includes:
- the laser radar in the working state in the laser radar system is determined.
- the determining the working lidar in the lidar system according to the scanning area range corresponding to the scene type and the scanning area range of each lidar in the lidar system including :
- the horizontal field of view angle of at least one of the lidars in the lidar system, the vertical angle of view of the lidar, and the lidar are determined according to the scene type.
- a scenario parameter adjustment strategy for one or more of the optical axis direction of the emitted beam, the scanning density of the lidar, the scanning frequency of the lidar, and the pulse transmission power of the lidar including:
- the identifier of at least one of the lidars in the lidar system, and the scene parameter adjustment strategy determine the scene type and the at least one laser in the lidar system.
- the identifier of the radar corresponds to the horizontal field of view of at least one lidar in the lidar system, the vertical angle of view of the lidar, the optical axis direction of the emitted beam of the lidar, and the scanning density of the lidar , a scene parameter adjustment strategy for one or more of the scanning frequency of the lidar and the pulse transmit power of the lidar.
- the scene parameter adjustment strategy of at least one of the lidar is adjusted according to the drivable area, and the adjusted scene parameter adjustment strategy is determined as at least one of the A parameter adjustment strategy of the laser radar, and based on the parameter adjustment strategy to adjust at least one current working parameter of the laser radar, specifically including:
- the scene type obtain the standard driving area under the scene type and the driving parameter adjustment of the horizontal field of view and the vertical field of view of at least one of the lidar systems in the lidar system corresponding to the standard driving area Strategy;
- the current working parameters of at least one of the lidars in the lidar system are adjusted based on the scene parameter adjustment strategy of at least one of the lidars in the lidar system.
- the standard driving area and the drivable area are matched, and according to the matching result and the driving parameter adjustment strategy of at least one of the lidars in the lidar system, the A scene parameter adjustment strategy for at least one of the lidars in the lidar system, including:
- the scene parameter adjustment strategy of at least one of the lidars in the lidar system is adjusted according to the scene difference coefficient and the driving parameter adjustment strategy of at least one of the lidars in the lidar system.
- the radar's scene parameter adjustment strategy includes:
- the vertical field of view and the horizontal field of view in the scene parameter adjustment strategy of at least one of the lidars in the lidar system are adjusted.
- a lidar system including a plurality of lidars, and the plurality of lidars are arranged on the same plane of the lidar system carrying device, or are arranged on the lidar system On different planes of the carrying device, the lidar includes: a transmitting device, a receiving device, a processor, a memory, a communication interface and a communication bus, and the processor, the memory and the communication interface complete each other through the communication bus. communication between;
- the transmitting device is used for transmitting outgoing laser light to the detection area
- the receiving device is used for receiving the echo laser reflected by the object in the detection area
- the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to perform the steps according to the above-mentioned method for adjusting the parameters of the laser radar to adjust the working parameters of the transmitting device.
- a computer storage medium where at least one executable instruction is stored in the storage medium, and the executable instruction causes the processor to execute the method for adjusting the parameters of the laser radar as described above. A step of.
- the embodiment of the present invention acquires the three-dimensional environment information around the lidar, identifies the scene type and drivable area where the lidar is located based on the 3D environment information, determines the parameter adjustment strategy of the lidar according to the scene type and the drivable area, and adjusts the parameters based on the parameter adjustment
- the strategy adjusts the current working parameters of the lidar, which can automatically adjust the working parameters of the lidar according to different scenarios, so that the working state of the lidar is automatically adjusted to the working state suitable for the current scene, thereby improving the working efficiency of the lidar, and can also Real-time identification of the current scene and detection of the drivable area, so as to adjust the scanning range of the lidar to match the drivable area, not only improves the obstacle detection rate, but also improves the utilization of computing resources and reduces power consumption.
- FIG. 1 shows a schematic structural diagram of an application scenario provided by an embodiment of the present invention
- FIG. 2 shows a schematic flowchart of a method for adjusting a laser radar parameter according to an embodiment of the present invention
- Figures 3a-3c show point cloud feature maps at different stages of data processing in an embodiment of the present invention
- FIG. 4 shows a schematic flowchart of neural network training in an embodiment of the present invention
- Fig. 5a shows a schematic diagram of the optical path for reducing the vertical field of view of the emitted light beam of the lidar in an embodiment of the present invention
- Fig. 5b shows a schematic diagram of the optical path for reducing the optical axis direction of the emitted light beam of the lidar in an embodiment of the present invention
- Fig. 6a and Fig. 6b are schematic diagrams showing the topology of a drivable area provided by an embodiment of the present invention.
- FIG. 7 shows a schematic diagram of the detection range of the lidar before and after parameter adjustment provided by an embodiment of the present invention
- FIG. 8a shows a schematic structural diagram of an application scenario provided by another embodiment of the present invention.
- FIG. 8b shows a schematic diagram of an application scenario including three lidars provided by another embodiment of the present invention.
- FIG. 8c shows a schematic diagram of an application scenario including five lidars provided by another embodiment of the present invention.
- FIG. 9 shows a schematic flowchart of a method for adjusting a lidar parameter provided by another embodiment of the present invention.
- FIG. 10 shows a schematic structural diagram of a lidar parameter adjustment device provided by an embodiment of the present invention.
- FIG. 11 shows a schematic structural diagram of a laser radar according to an embodiment of the present invention.
- FIG. 1 shows a schematic diagram of an application scenario provided by an embodiment of the present invention.
- the application scenario includes: a lidar 101 and a vehicle 102 .
- the lidar 101 is installed on various vehicles 102 that need to detect the surrounding environment, and the vehicle 102 can be, for example, a vehicle, a ship, an aircraft, or the like.
- the lidar when the vehicle 102 is a vehicle, the lidar can be installed on the front, rear, roof, side of the vehicle or any other body position where the lidar can be fixed. It can be understood that the lidar can also be fixed on Other devices that are externally connected to the vehicle are on the structural platform. It can be understood that the external device structure platform can keep synchronized with the vehicle when the vehicle is running.
- FIG. 2 shows a schematic flowchart of a method for adjusting parameters of a laser radar according to an embodiment of the present invention. This method is applied to the lidar 101 in FIG. 1 . As shown in Figure 2, the method includes:
- Step 110 Acquire three-dimensional environment information around the lidar.
- the lidar may be a rotating mechanical lidar or a solid-state lidar. It is understandable that when the lidar is a rotating mechanical lidar, the 3D environment information covers a 360-degree field of view around the lidar. When the lidar is a solid-state lidar, the 3D environment information covers a certain angle range in front of the lidar, for example is 120 degrees.
- the three-dimensional environment information may be three-dimensional point cloud data.
- the three-dimensional point cloud data may include geometric position information and reflection intensity information.
- the reflection intensity information is the echo intensity collected by the laser radar receiving device, and the intensity information is related to the surface material, roughness, incident angle direction of the target, and the emitted energy and laser wavelength of the instrument.
- the three-dimensional environment information may also be the combined data of three-dimensional point cloud data and image data, obtaining a color image through an image sensor, and then assigning the color information (RGB) of the pixel at the corresponding position to the data.
- the corresponding points in the point cloud so that the 3D point cloud data has color information.
- the image sensor can be integrated inside the lidar, so that the acquisition of the color information corresponding to the point cloud data is completed before the point cloud data is output; so as to identify the semantic features of the three-dimensional environment; optionally, the image sensor can also be independent It is installed outside the radar and integrates point cloud data and image data into three-dimensional environment information.
- the image data may be image data or image data.
- the three-dimensional environment information around the lidar is acquired in real time.
- Step 120 based on the three-dimensional environment information, identify the scene type and the drivable area where the lidar is located.
- the type of scene in which the lidar is located is related to the field in which the lidar is used.
- the scene in which it is located is also different, so the scene types can also include multiple types.
- the scene types include but are not limited to: highway scene types, intersection scene types, parking lot scene types, and the like.
- Step 121 Process the three-dimensional point cloud data to generate a multi-channel point cloud feature map
- the three-dimensional point cloud data is processed to generate a multi-channel point cloud feature map, which specifically includes:
- each channel in the multi-channel feature map represents a feature value of a certain dimension.
- the characteristic values corresponding to multiple channels may include but are not limited to the following: occupancy rate, number of points, height of the lowest point, height of the highest point, average height, average intensity, distance, azimuth, etc.
- the grid size can be determined to be 0.2 meters, and the front, back, left, and right ranges can be determined to be 80 meters each to form a feature map with a size of 800 ⁇ 800.
- selecting the occupancy rate, the number of points, the average height and the average intensity of the four features will constitute an 800 ⁇ 800 ⁇ 4 point cloud feature map.
- Step 122 extracting high-dimensional feature information from the multi-channel point cloud feature map
- extracting high-dimensional feature information from a multi-channel point cloud feature map specifically includes: inputting the multi-channel point cloud feature map into a first neural network, and the first neural network obtains high-dimensional feature information through a convolution operation, wherein the The high-dimensional feature information can include: edge feature information, shape feature information, texture feature information, and even semantic features, such as vehicle outline information, toll booth outline information, traffic light outline information, road traffic sign information, road Along the information, lane line information, pedestrian crossing information, etc. It can be understood that when the three-dimensional environment information is the combined data of three-dimensional point cloud data and image data, the high-dimensional feature information extracted from the multi-channel point cloud feature map has color features, which can be used for road signage. , traffic lights and other semantic features with text or color recognition.
- Step 123 Determine the scene type according to the high-dimensional feature information
- the determining the scene type according to the high-dimensional feature information specifically includes: inputting the high-dimensional feature information into a second neural network to obtain an output value of the scene type, and according to the difference between the scene type output value and the scene type label Corresponding relationship, the scene type output value is matched with the scene type label, the scene type corresponding to the scene type output value is determined, and the scene type in which the lidar is located is determined.
- the scene type label includes a scene type output value range with the scene type label.
- the matching of the scene type output value with the scene type label according to the corresponding relationship between the preset scene type output value and the scene type label specifically includes: matching the scene type output value with the scene type label. When the scene type output value falls within the scene type output value range corresponding to the scene type label, it is determined that the scene type is the scene type corresponding to the scene type label.
- the obtained scene output value is 9.
- the scene type corresponding to the intersection scene label is output The value range is 7-10, then the label corresponding to the scene is the intersection scene; for example, if the high-dimensional feature is the crosswalk zebra crossing, the traffic light outline information, the scene type output value obtained by inputting the second neural network is 8, the scene also falls into the scene output value range corresponding to the intersection scene label, and the label corresponding to this scene is the intersection scene.
- the scene type in which the lidar is located is determined by the corresponding relationship between the scene type output value and the scene type label, which increases the accuracy of the scene type judgment.
- the inputting the high-dimensional feature information into the second neural network to obtain the scene type output value may specifically include: inputting the high-dimensional feature information into the second neural network, and inputting the high-dimensional feature information into the second neural network.
- the high-dimensional feature information is obtained by matching the feature information in the second neural network model library, wherein it can be understood that the feature information in the second neural network model library has eigenvalues and weight values, and according to the matched eigenvalues and the weight value to obtain the scene type output value.
- Step 124 Determine a drivable area according to the high-dimensional feature information.
- determining the drivable area according to the high-dimensional feature information specifically includes: inputting the high-dimensional feature information into a third neural network to obtain an output map of the drivable area, and outputting the map and the drivable area according to a preset drivable area The corresponding relationship between the labels is to match the output map of the drivable area with the labels of the drivable area to determine the drivable area corresponding to the output map of the drivable area, so as to determine the drivable area.
- inputting the high-dimensional feature information into the third neural network to obtain the output map of the drivable area may include: inputting the high-dimensional feature information into the third neural network, and the third neural network performs drivable operation.
- Area feature extraction wherein the drivable area features may include, for example, road edge information, lane line information, etc.; extracting the area containing the drivable area features to generate an output map of the drivable area.
- the drivable area labels in different scenarios may be different.
- the drivable area label of a highway may be a label containing lane line information
- the drivable area label of an intersection may include road edges.
- Information and lane line information, the drivable area label of the parking lot can include parking space line information and so on.
- Matching the output map of the drivable area with the label of the drivable area to determine the drivable area may include: matching the output map of the drivable area with the label of the drivable area, and determining the scene in which the output map of the drivable area is located. The above scenario determines the drivable area.
- the method further includes: acquiring obstacle information in the drivable area, and determining the drivable area according to the obstacle information If the complexity of the drivable area exceeds a preset threshold, re-acquire the drivable area.
- the first neural network and the second neural network are both composed of convolution layers and pooling layers
- the third neural network is composed of deconvolution layers.
- the high-dimensional feature information is an intermediate variable, which is obtained by the first neural network through the convolution operation, and is input into the second neural network and the third neural network respectively.
- Figures 3a-3c are point cloud feature maps at different processing stages, in which Figure 3a is the point cloud feature map (original 3D point cloud data) in the early stage of data processing, and Figure 3b is the point cloud in the middle stage of data processing Feature map (output after passing through the first neural network), Figure 3c is the point cloud feature map after data processing (output after passing through the second neural network).
- nonlinear activation function since the basic operation of the neural network is a convolution operation, which consists of a series of multiplication and addition operations, in order to introduce nonlinear transformation, a nonlinear activation function will be added to the neural network.
- Commonly used nonlinear activation functions are ReLU, Sigmoid, etc.
- the method before step 120, further includes: acquiring training point cloud data, where the training point cloud data includes original point cloud data and a scene type corresponding to the original point cloud data. label and drivable area label; according to the original point cloud data and the scene type label and drivable area label corresponding to the original point cloud data, the first neural network, the second neural network and the third neural network are trained.
- the original point cloud data can be obtained by vehicle lidar, and the scene type label and drivable area label corresponding to the original point cloud data can be obtained by manual annotation.
- the larger the amount of original point cloud data in the training set formed by the original point cloud data the more accurate the neural network obtained by training.
- the training of the first neural network, the second neural network, and the third neural network may specifically include: respectively acquiring the initialization parameters of the first neural network, the second neural network, and the third neural network; Generate a training point cloud feature map from the original point cloud data in the training point cloud data; input the training point cloud feature map into the first neural network, the second neural network, and the third neural network, and obtain the training scene output by the second neural network type, the training area of interest output by the third neural network; compare the training scene type with the scene type label, and calculate the scene type loss value through the second loss function; compare the training area of interest with the drivable area label, through the third loss The function calculates the loss value of the drivable area; based on the loss value of the scene type and the loss value of the drivable area, the parameters of the first neural network, the second neural network and the third neural network are updated through the back-propagation algorithm; Stop training when the loss value of the drivable area no longer changes, and save
- Step 130 Determine the parameter adjustment strategy of the lidar according to the scene type and the drivable area, and adjust the current working parameters of the lidar based on the parameter adjustment strategy.
- the current working parameters of the laser radar include, but are not limited to: the vertical field of view of the laser radar's emission beam, the horizontal field of view of the laser radar's emission beam, and the optical axis direction of the laser radar's emission beam , one or more of the scanning density of the lidar, the scanning frequency of the lidar, and the pulse transmission power of the lidar.
- step 130 may specifically include:
- Step 131 Determine the horizontal field of view angle of the lidar, the vertical field of view of the lidar, the optical axis direction of the emitted beam of the lidar, the scanning density of the lidar, the scanning frequency of the lidar, and the scanning frequency of the lidar according to the scene type.
- a scenario parameter adjustment strategy for one or more of the pulse transmit power is provided.
- the scenario parameter adjustment strategy and the corresponding relationship between the scenario type and the scenario parameter adjustment strategy are preset by the user according to the scenario type.
- the lidar when a vehicle loaded with lidar is driving in a highway scene, the lidar is required to focus the scanning range on the front and rear road areas, reducing the detection of the left and right sides of the vehicle, and ignoring or reducing the detection outside the drivable area. Reducing the detection range not only saves computing power, but also improves the detection rate of obstacles in key areas. Therefore, in some optional embodiments, when a highway scene type is detected, the vertical field of view of the lidar is decreased, the pulse emission frequency of the emitted beam of the lidar is decreased, and the scanning frequency of the lidar is increased.
- the vertical field of view of the emitted beam can be compressed by 20%, or the vertical field of view of the lidar can be adjusted from +15° ⁇ -25° to +12° ⁇ -20°, can concentrate more wire beams in the horizontal direction, increase the front and rear detection distance of the lidar, and thus detect long-distance targets.
- reducing the pulse emission frequency such as adjusting the pulse emission frequency from 50,000 per second to 10,000 per second, can increase the charging time of the pulse laser transmitter, increase the laser energy, and detect farther distance.
- increasing the scanning frequency such as adjusting the scanning frequency from 10Hz to 15Hz, can detect changes in moving objects faster, so as to improve the safety of autonomous driving.
- the pulse emission frequency of the lidar is increased, the optical axis direction of the emitted beam of the lidar is decreased, and the scanning frequency of the lidar is increased.
- the pulse emission frequency of lidar is increased, for example, the pulse emission frequency is adjusted from 10,000 pulses per second to 50,000 pulses per second, so that changes in surrounding moving objects can be obtained faster.
- reduce the optical axis direction of the emitted beam as shown in Figure 5b, for example, the optical axis direction of the emitted beam can be moved down by 20°. With a downward deflection of 20°, the detection range of the emitted beam will be closer to the ground, which can reduce the detection distance of the lidar, so as to focus on detecting short-range targets.
- the scanning frequency of lidar can be increased, for example, the scanning frequency can be adjusted from 10Hz to 15Hz, which can detect changes in moving objects faster.
- the power consumption of lidar can be reduced while the surrounding is fixed. Therefore, in some optional embodiments, when a parking lot scene type is detected, the pulse emission frequency of the lidar is decreased, the optical axis direction of the emitted beam of the lidar is decreased, and the scanning frequency of the lidar is decreased. Among them, reducing the scanning frequency of the lidar, for example, the scanning frequency can be adjusted from 15Hz to 10Hz, which can reduce the power consumption of the lidar.
- Step 132 Adjust the scene parameter adjustment strategy according to the drivable area, determine the adjusted scene parameter adjustment strategy as the lidar parameter adjustment strategy, and adjust the parameter adjustment strategy based on the parameter adjustment strategy.
- the current operating parameters of the lidar are the parameters that are adjusted.
- step 132 may specifically include:
- Step 1321 According to the scene type, obtain the standard driving area under the scene type and the driving parameter adjustment of the horizontal field of view of the lidar and the vertical field of view of the lidar corresponding to the standard driving area Strategy.
- the corresponding relationship between the scene type and the standard driving area is preset, and then the standard driving area under the current scene type can be searched from the corresponding relationship between the scene type and the standard driving area according to the current scene type. For example, if the standard driving area corresponding to the highway scene type is preset as a specific straight road area, then when the scene type is determined to be the expressway scene type, the corresponding standard driving area is found to be a specific straight road area. At the same time, there is a corresponding relationship between the scene type, the standard driving area, and the driving parameter adjustment strategy. In an ideal situation, the drivable area of a certain scene type should be the standard driving area, and the preset standard field of view should be set according to the standard driving area.
- the actually determined drivable area is not necessarily exactly the same as the standard driving area. Therefore, the standard driving area and the horizontal field of view of the lidar and the vertical field of view of the lidar are also preset.
- the corresponding relationship of the parameter adjustment strategy can be found in the corresponding relationship between the standard driving region and the driving parameter adjustment strategy of the horizontal field of view of the lidar and the vertical field of view of the lidar according to the obtained standard driving area.
- the driving parameter adjustment strategy of the horizontal field of view of the lidar and the vertical field of view of the lidar corresponding to the acquired standard driving area.
- Step 1322 Match the drivable area with the standard driving area, and readjust the scene parameter adjustment strategy according to the matching result and the driving parameter adjustment strategy.
- the standard driving area and the drivable area are matched, and the scene parameter adjustment strategy is adjusted according to the matching result and the driving parameter adjustment strategy, which specifically includes : Analyze the topological structure of the drivable area and the topological structure of the standard driving area, and calculate the topological correlation between the topological structure of the drivable area and the topological structure of the standard driving area; The scene difference coefficient of the driving area; wherein, the scene difference coefficient includes translation difference coefficient, rotation difference coefficient, scaling difference coefficient, etc.; according to the scene difference coefficient and the driving parameter adjustment strategy, adjust the scene parameter adjustment strategy .
- adjusting the scene parameter adjustment strategy according to the scene difference coefficient and the driving parameter adjustment strategy specifically includes: acquiring standard field of view parameters (including standard vertical viewing angle parameters) in the driving parameter adjustment strategy. Field angle parameters and standard horizontal field angle parameters); according to the scene difference coefficient, new field angle parameters (including new vertical field angle and new horizontal field angle are calculated proportionally on the basis of standard field angle parameters) parameter); according to the new field of view parameters, adjust the vertical field of view and horizontal field of view in the scene parameter adjustment strategy.
- topological structure of the drivable area and the topology of the standard driving area have coincident centers, parallel sides, the same width, and a length ratio of 5:3, then determine the topology of the drivable area and the standard
- the topological structure of the driving area has a translation difference coefficient of 0, a rotation difference coefficient of 0, and a scaling difference coefficient of (5/3)*1.
- FIG. 6b assuming that the topological structure of the drivable area and the topological structure of the standard driving area have the same centers, parallel sides, a width ratio of 6:4, and a length ratio of 6:4, it is determined that the drivable area is drivable.
- the translation difference coefficient of the topology of the area and the topology of the standard driving area is 0, the rotation difference coefficient is 0, and the scaling difference coefficient is (6/4)*(6/4).
- the topology of the standard driving area can be kept unchanged, and the drivable area Rotate and translate the topology of the drivable area until the centers of the topology of the drivable area and the topology of the standard driving area are coincident and the sides are parallel, and record the translation difference coefficient and rotation difference coefficient, and then calculate the topology of the drivable area and the standard
- the proportional relationship between the length and width of the topology of the driving area determines the scaling difference factor.
- a new field angle parameter may be calculated according to the translation difference coefficient, the rotation difference coefficient, and the scaling difference coefficient.
- the magnification of the field of view angle can be calculated according to the translation difference coefficient, rotation difference coefficient, zoom difference coefficient, and the proportional relationship of the sides of similar graphics.
- the optical axis direction of the emitted beam of the lidar can be adjusted according to the rotational difference coefficient, so that the optical axis direction of the adjusted laser beam emitted by the lidar is parallel to the standard.
- the central axis of the topology structure of the driving area and then according to the translation difference coefficient, the scaling difference coefficient, and the proportional relationship of the edges of similar graphics, the magnification of the field of view angle is calculated, and the new field of view angle parameters are calculated to make
- the detection range of the lidar can completely cover the standard driving area.
- the difference coefficient between the new detection range and the drivable area will be calculated again according to the above method, and the field angle parameters (vertical field angle and horizontal field angle parameters) will be further adjusted on this basis. .
- Step 1323 Adjust the current working parameters of the lidar based on the adjusted scene parameter adjustment strategy.
- the parameter adjustment range or the preset adjusted parameter value cannot exceed the executable range of the lidar.
- the monitoring area is prompted to be readjusted.
- determining a scene parameter adjustment strategy that matches the highway scene type includes: compressing the vertical field of view of the laser radar's emission beam from +45° to 40°, and reducing the pulse emission frequency of the laser radar's emission beam from the emission per second Adjust from 50,000 to 10,000 per second, adjust the scanning frequency of lidar from 10Hz to 15Hz, and adjust the driving parameter adjustment strategy to match the type of highway scene: the horizontal field of view is 120°, the vertical field of view is 120° is 35°, assuming that the topological correlation between the topological structure of the drivable area and the topological structure of the standard driving area is that the center coincides, the sides are parallel, the width is the same, and the length ratio is 3:5, that is, the translation difference coefficient is 0, and the rotation difference coefficient is 0.
- determining the scene parameter adjustment strategy that matches the scene type of the intersection includes: adjusting the pulse emission frequency from 10,000 pulses per second to 50,000 pulses per second, moving the optical axis direction of the emitted beam down by 20°, and scanning The frequency is adjusted from 10Hz to 15Hz, and the driving parameter adjustment strategy that matches the scene type of the intersection includes: the horizontal field of view is 120°, and the vertical field of view is 90°, assuming that the topology of the drivable area is the same as that of the standard driving area.
- the adjusted scene parameter adjustment strategy includes: adjusting the horizontal field of view of the laser radar’s emission beam to 180°, adjusting the vertical field of view of the lidar’s emission beam to 135°, and changing the pulse emission frequency from 1 per second. Adjust the 10,000 to emit 50,000 per second, move the optical axis direction of the emitted beam down by 20°, and adjust the scanning frequency from 10Hz to 15Hz.
- determining the scene parameter adjustment strategy that matches the scene type of the parking lot includes: adjusting the pulse emission frequency of the emission beam of the lidar from 50,000 pulses per second to 10,000 pulses per second, adjusting the pulse emission frequency of the emission beam of the lidar
- the axis direction is reduced by 20°
- the scanning frequency is reduced from 15Hz to 10Hz
- the driving parameter adjustment strategy that matches the parking lot scene type includes: the horizontal field of view is 120°, and the vertical field of view is 60°.
- the topological correlation between the topological structure and the topological structure of the standard driving area is that the center is coincident, the sides are parallel, the ratio of width is 1:2, and the ratio of length is 1:2, that is, the translation difference coefficient is 0, the rotation difference coefficient is 0, and the scaling difference is 0.
- the adjusted scene parameter adjustment strategy includes: adjusting the horizontal field of view of the laser radar beam to 60°, adjusting the vertical field of view of the laser beam beam to 30°, and adjusting the laser radar beam
- the pulse emission frequency of the emitted beam of the laser radar is adjusted from 50,000 to 10,000 per second, the optical axis direction of the laser beam is reduced by 20°, and the scanning frequency is reduced from 15Hz to 10Hz.
- the embodiment of the present invention obtains the three-dimensional environment information around the laser radar, identifies the scene type and the drivable area where the laser radar is located based on the three-dimensional environment information, determines the parameter adjustment strategy of the laser radar according to the scene type and the drivable area, and determines the parameter adjustment strategy of the laser radar based on the parameter
- the adjustment strategy adjusts the current working parameters of the lidar, which can automatically adjust the working parameters of the lidar according to different scenarios, so that the working state of the lidar is automatically adjusted to the working state suitable for the current scene, thereby improving the working efficiency of the lidar, and also It can identify the current scene and detect the drivable area in real time, so as to adjust the scanning range of the lidar to match the drivable area, which can not only improve the obstacle detection rate, but also improve the utilization of computing resources and reduce power consumption.
- FIG. 8a shows a schematic structural diagram of an application scenario provided by another embodiment of the present invention.
- the application scenario includes: a lidar system 103 and a vehicle 102 .
- the lidar system 103 is installed on various vehicles 102 that need to detect the surrounding environment, and the vehicles 102 may be vehicles, ships, aircrafts, and the like.
- the lidar system 103 includes a plurality of lidars. Multiple lidars can be placed on the same plane or on different planes. When the vehicle 102 is a vehicle, the lidar can be installed at one or more places in the front, rear, and roof of the vehicle. For example, as shown in FIG.
- the lidar system 103 includes three lidars, and the three lidars 1031, 1032, and 1033 are all arranged on the roof of the vehicle, and the lidar 1031 is located at the center of the roof, and the lidars 1032, 1032, 1033 is located on the opposite sides of the roof, the lidar 1031 can be used to obtain the environmental information around the vehicle and far away from the vehicle, and the lidars 1032 and 1033 are respectively used to obtain the environmental information on the left and right sides of the vehicle and are closer to the vehicle; For example, as shown in FIG.
- the lidar system 103 includes five lidars, the lidar 1034 is arranged on the roof of the vehicle and is located in the center of the roof, and the lidars 1035 and 1036 are arranged on the left and right sides of the vehicle and are located in the center of the roof.
- the lidar 1037 sets the front of the vehicle
- the lidar 1038 sets the rear of the vehicle.
- FIG. 9 shows a schematic flowchart of a method for adjusting a lidar parameter provided by another embodiment of the present invention.
- the method is applied to the lidar system 103 in Figures 8a to 8c. As shown in Figure 9, the method includes:
- Step 210 Acquire three-dimensional environment information around the lidar system.
- step 210 is substantially the same as that of step 110 in the foregoing embodiment.
- the three-dimensional environment information around the laser radar system is obtained through one laser radar in the laser radar system, and the three-dimensional environment information around the laser radar system can also be obtained through multiple laser radars in the laser radar system. For example, as shown in FIG.
- the details may be: obtaining the three-dimensional environment around the lidar system 103 through the lidar 1031 in the lidar system 103 information; when the three-dimensional environment information around the lidar system 103 is acquired through multiple lidars in the lidar system, the details can also be: part of the three-dimensional environment information ( For example, the lidar 1031 obtains the front 3D environment information, the lidar 1032 obtains the right 3D environment information, and the lidar 1033 obtains the left 3D environment information), and integrates the obtained 3D environment information to obtain complete 3D environment information.
- Step 220 based on the three-dimensional environment information, identify the scene type and the drivable area where the lidar system is located.
- step 220 is substantially the same as that of step 120 in the foregoing embodiment, and details are not described herein again.
- Step 230 Determine the parameter adjustment strategy of each lidar in the lidar system according to the scene type and the drivable area, and adjust the current working parameters of each lidar respectively based on the parameter adjustment strategy of each lidar.
- step 230 includes:
- Step 231 Determine one or more of the horizontal field of view, vertical field of view, optical axis direction of the emitted beam, scanning density, scanning frequency, and pulse transmission power of each lidar in the lidar system according to the scene type. Scene parameter adjustment strategy.
- the scenario parameter adjustment strategy and the corresponding relationship between the scenario type and the scenario parameter adjustment strategy are preset by the user according to the scenario type.
- the scene parameter adjustment strategies of different lidars in the lidar system can be different.
- the corresponding relationship between the scene type and the scene parameter adjustment strategies of different lidars can also be preset by the user according to the scene types and the positions of different lidars.
- one or more of the horizontal field of view, vertical field of view, optical axis direction of the emitted beam, scanning density, scanning frequency, and pulse emission power of each lidar in the lidar system is determined according to the scene type.
- a variety of scene parameter adjustment strategies specifically: obtaining the lidar identification in the lidar system, determining and identifying the scene type and the corresponding relationship between the preset scene type, lidar identification and scene parameter adjustment strategy The scene parameter adjustment strategy corresponding to the acquired lidar identifier.
- the identifiers of the lidars 1031, 1032, and 1033 are a, b, and c, respectively, determine the scene parameter adjustment strategy corresponding to the identified scene type and the identifiers a, b, and c.
- the scene parameter adjustment strategy corresponding to a as follows: reduce the vertical field of view of the lidar, reduce the pulse emission frequency of the emitted beam of the lidar, and increase the scanning frequency of the lidar, determine
- the adjustment strategies for scene parameters corresponding to b and c are: reducing the horizontal field of view of the lidar, reducing the pulse emission frequency of the laser beam emitted by the lidar, and increasing the scanning frequency of the lidar, thereby increasing the front and rear detection of the lidar on the roof Distance, compress the left and right detection distance of the lidar on both sides of the roof, and detect the change of moving objects faster; when the current scene type is detected as the intersection scene type, determine the scene parameter adjustment strategy corresponding to a as follows: Increase the vertical field of view of the lidar, increase the pulse emission frequency of the lidar, reduce the optical axis direction of the emitted beam of the lidar, and increase the scanning frequency of the lidar.
- the strategy is: reduce the vertical field of view of the lidar, reduce the pulse emission frequency of the lidar and reduce the scanning frequency of the lidar, and determine the scene parameter adjustment strategy corresponding to b and c as follows: reduce the horizontal field of view of the lidar, Reduce the pulse emission frequency of the lidar and reduce the scanning frequency of the lidar, thereby reducing the detection range of the lidar and reducing the power consumption of the lidar.
- the parameter adjustment strategy of each lidar may be determined according to the scanning area range of each lidar.
- the standard driving area and the drivable area are matched, and according to the matching result and the driving parameter adjustment strategy of each lidar in the lidar system, the scene parameter adjustment strategy of each lidar in the lidar system is adjusted respectively.
- the method further includes: acquiring the scanning area range of each laser radar in the laser radar system; and determining a parameter adjustment strategy of each laser radar according to the scanning area range of each laser radar.
- the center lidar 1031 is used for detecting the medium and long-distance field of view
- the lidar 1033 and the lidar 1032 are respectively used for short-range field of view detection and left and right field of view detection;
- the central axis of the lidar 1031 is perpendicular to the horizontal plane; wherein, the central axis of the lidar 1031 is the bottom surface normal passing through the center of the bottom surface of the lidar 1031 .
- the lidar 1031 is located on the left side of the central lidar, and the angle between the central axis of the lidar 1033 and the central axis of the lidar 1031 is less than zero degrees; the lidar 1032 is located on the right side of the lidar, and the central axis of the lidar 1032 The included angle with the central axis of the lidar 1031 is greater than zero degrees; wherein, the central axis of the lidar 1032 and the lidar 1033 is the bottom normal line passing through the center of the bottom surface of the lidar.
- each lidar in the lidar system may be on the same plane or on different planes.
- the detection requirement of the medium and far field of view needs to be strengthened according to the identified scene, one or more of the scanning frequency, scanning density, and pulse transmission power of the lidar 1031 are adjusted.
- the optical axis direction, scanning density, scanning frequency and One or more of the pulsed transmit power are adjusted.
- the lidar system can also select part of the lidar work or all lidar work according to the identified scene; For the detection of medium and far field of view), the laser radar 1031 can be selected to work, and the adjustment parameters of the laser radar 1031 can be determined according to the scene and the drivable area. When it is identified as an intersection scene (that is, the detection of objects around the vehicle needs to be enhanced), all lidars in the system can be turned on at the same time, and the working parameters of lidar 1033 and lidar 1032 in the system (as shown in Figure 8b) can be adjusted.
- the lidar system 103 may also include five lidars, wherein the central lidar 1031 is used to detect the medium and long-distance field of view, and the lidar 1033 and lidar 1032 are respectively used In the left and right short field of view detection; 1034 and 1035 are short-range front and rear field of view detection.
- the lidar system determines the parameter adjustment strategy of each lidar according to the scanning area range of each lidar. Specifically, for example, as shown in FIG. 8c , when it is identified as a highway scene (that is, the detection of the medium and far field of view is more necessary), the lidar 1031 can be selected to work, and the adjustment parameters of the lidar 1031 can be determined according to the scene and the drivable area.
- all lidars in the system can be turned on at the same time, and the lidar 1033, lidar 1032, lidar 1034 and lidar 1034 in the system (as shown in Figure 8b) can be adjusted Operating parameters of LiDAR 1035.
- Step 232 Adjust the scene parameter adjustment strategy of each lidar according to the drivable area, and determine the adjusted scene parameter adjustment strategy of each lidar as the parameter adjustment strategy of each lidar, and based on the parameter adjustment strategies of each lidar respectively. Adjust the current working parameters of each lidar.
- step 232 may specifically include:
- Step 2321 According to the scene type, obtain the standard driving area under the scene type and the driving parameter adjustment strategy of the horizontal field of view and vertical field of view of each lidar in the lidar system corresponding to the standard driving area.
- the corresponding relationship between the scene type and the standard driving area is preset, and then the standard driving area under the current scene type can be searched from the corresponding relationship between the scene type and the standard driving area according to the current scene type. For example, if the standard driving area corresponding to the highway scene type is preset as a specific straight road area, then when the scene type is determined to be the expressway scene type, the corresponding standard driving area is found to be a specific straight road area. At the same time, there is a corresponding relationship between the scene type, the standard driving area, and the driving parameter adjustment strategy. In an ideal situation, the drivable area of a certain scene type should be the standard driving area, and the preset standard field of view can be set according to the standard driving area.
- the actually determined drivable area is not necessarily exactly the same as the standard driving area. Therefore, it is also necessary to pre-set the standard driving area and the horizontal field of view and vertical field of view of each lidar system in the lidar system. Adjustment of the driving parameters According to the corresponding relationship of the strategy, the obtained standard driving area can be found in the corresponding relationship between the standard driving area and the driving parameter adjustment strategy of the horizontal field of view and vertical field of view of each lidar in the lidar system according to the acquired standard driving area.
- the driving parameter adjustment strategy for the horizontal field of view and vertical field of view of each lidar corresponding to the area.
- Step 2322 Match the standard driving area and the drivable area, and adjust the scene parameter adjustment strategy of each lidar in the lidar system according to the matching result and the driving parameter adjustment strategy of each lidar in the lidar system.
- the standard driving area and the drivable area are matched, and according to the matching result and the driving parameter adjustment strategy of each lidar in the lidar system, the driving parameters of each lidar in the lidar system are adjusted respectively.
- the scene parameter adjustment strategy specifically includes: analyzing the topological structure of the drivable area and the topological structure of the standard driving area, and calculating the topological correlation between the topological structure of the drivable area and the topological structure of the standard driving area; The scene difference coefficient of the driving area relative to the standard driving area, where the scene difference coefficient includes translation difference coefficient, rotation difference coefficient and scaling difference coefficient, etc.
- the scene parameter adjustment strategy of each lidar in the lidar system is adjusted respectively, which specifically includes: obtaining the data of a certain lidar.
- Standard view angle parameters in the driving parameter adjustment strategy including standard vertical view angle parameters and standard horizontal view angle parameters
- the new view angle parameters are calculated proportionally on the basis of the standard view angle parameters parameters (including new vertical field of view and new horizontal field of view parameters); according to the new field of view parameters, adjust the vertical field of view and horizontal field of view in the scene parameter adjustment strategy of the lidar; and follow the same way to adjust the scene parameter adjustment strategy of each lidar in the lidar system.
- Step 2323 adjusting the current working parameters of each lidar in the lidar system based on the scene parameter adjustment strategy of each lidar in the lidar system.
- the scene parameter adjustment strategy corresponding to a is: Compress the vertical field of view of the emission beam from +90° to 80°, adjust the pulse emission frequency of the emission beam from 50,000 pulses per second to 10,000 pulses per second, and adjust the scanning frequency from 10Hz to 15Hz.
- the scene parameter adjustment strategies corresponding to b and c are: compress the horizontal field of view of the emitted beam from +180° to 120°, and adjust the pulse emission frequency of the emitted beam from 50,000 pulses per second to 10,000 pulses per second , Adjust the scanning frequency from 10Hz to 15Hz; determine the driving parameter adjustment strategy corresponding to a as follows: the horizontal field of view is 120°, the vertical field angle is 60°, and the driving parameter adjustment strategies corresponding to b and c are: The field angle is 100°, and the vertical field angle is 35°; it is assumed that the topological correlation between the topology of the drivable area and the topology of the standard driving area is that the center is coincident, the sides are parallel, the width is the same, and the length ratio is 3:5.
- the scene parameter adjustment strategy is: adjust the horizontal field of view of the emission beam to 72°, adjust the vertical field of view of the emission beam to 60°, and adjust the pulse emission frequency of the emission beam from 50,000 pulses per second to 50,000 pulses per second. Send 10,000 pieces and adjust the scanning frequency from 10Hz to 15Hz.
- the adjusted scene parameter adjustment strategies corresponding to b and c are as follows: adjust the horizontal field of view of the emission beam to 60°, and adjust the vertical field of view of the emission beam. It is 35°, and the pulse emission frequency of the emission beam is adjusted from 50,000 to 10,000 per second, and the scan frequency is adjusted from 10 Hz to 15 Hz.
- the embodiment of the present invention obtains the three-dimensional environment information around the laser radar system, identifies the scene type and drivable area where the laser radar system is located based on the three-dimensional environment information, and determines the type of each laser radar in the laser radar system according to the scene type and the drivable area.
- Parameter adjustment strategy and adjust the current working parameters of each lidar based on the parameter adjustment strategy of each lidar, which can automatically adjust the working parameters of each lidar in the lidar system according to different scenarios, so that the working state of the lidar system can be automatically adjusted.
- FIG. 10 shows a schematic structural diagram of a lidar parameter adjustment apparatus provided by an embodiment of the present invention.
- the device can be applied to the lidar or lidar system in the above embodiments.
- the apparatus 600 includes: an environment information acquisition module 610 , an identification module 620 and an adjustment module 630 .
- the environmental information acquisition module 610 is used to acquire the three-dimensional environmental information around the lidar
- the identification module 620 is used to identify the scene type and drivable area where the lidar is located based on the three-dimensional environment information; the adjustment module 630 is used to determine the parameter adjustment strategy of the lidar according to the scene type and the drivable area, and adjust based on the parameter adjustment strategy The current operating parameters of the lidar.
- the 3D environment information includes 3D point cloud data; the identification module 620 is specifically configured to: process the 3D point cloud data to generate a multi-channel point cloud feature map; High-dimensional feature information is extracted from the point cloud feature map; the scene type is determined according to the high-dimensional feature information; and the drivable area is determined according to the high-dimensional feature information.
- the identification module 620 is specifically configured to: input the multi-channel point cloud feature map into a first neural network, and obtain the high-dimensional feature information output by the first neural network;
- the high-dimensional feature information is input into the second neural network, and the scene type output value output by the second neural network is obtained; according to the corresponding relationship between the preset scene type output value and the scene type label, determine the corresponding scene type output value.
- the scene type input the high-dimensional feature information into the third neural network, and obtain the output map of the drivable area output by the third neural network; according to the preset corresponding relationship between the output map of the drivable area and the label of the drivable area , and determine the drivable area corresponding to the drivable area output map.
- the adjustment module 630 is specifically configured to: determine the horizontal field of view of the lidar, the vertical angle of view of the lidar, and the emission beam of the lidar according to the scene type One or more scene parameter adjustment strategies among the optical axis direction of the laser radar, the scanning density of the laser radar, the scanning frequency of the laser radar, and the pulse transmission power of the laser radar; according to the drivable area, adjust The scene parameter adjustment strategy is determined, and the adjusted scene parameter adjustment strategy is determined as the parameter adjustment strategy of the lidar, and the current working parameters of the lidar are adjusted based on the parameter adjustment strategy.
- the adjustment module 630 is further configured to: obtain, according to the scene type, a standard driving area under the scene type and the horizontal field of view of the lidar corresponding to the standard driving area and the driving parameter adjustment strategy of the vertical field of view of the lidar; the standard driving area and the drivable area are matched, and the scene parameters are adjusted according to the matching result and the driving parameter adjustment strategy Adjusting the strategy; adjusting the current working parameters of the lidar based on the adjusted scene parameter adjustment strategy.
- the adjustment module 630 is further configured to: analyze the topology structure of the drivable area and the topology structure of the standard driving area; calculate the topology structure of the drivable area and the standard driving area The topological correlation of the topological structure of the area; according to the topological correlation, determine the scene difference coefficient of the drivable area relative to the standard driving area; according to the scene difference coefficient and the driving parameter adjustment strategy, adjust the Describe the scene parameter adjustment strategy.
- the current working parameters of the lidar include: the vertical field of view angle of the emitted beam of the lidar, the horizontal field of view of the emitted beam of the lidar, the One or more of the optical axis direction of the emitted light beam, the scanning frequency of the lidar, and the pulse emission frequency of the lidar.
- the laser radar parameter adjustment device provided by the embodiment of the present invention is a device capable of executing the above-mentioned laser radar parameter adjustment method, and all the above-mentioned embodiments based on the laser radar parameter adjustment method are applicable to the device, and can achieve same or similar beneficial effects.
- the three-dimensional environment information around the lidar is obtained, based on the three-dimensional environment information, the scene type and the drivable area where the lidar is located is identified, and the parameter adjustment strategy of the lidar is determined according to the scene type and the drivable area, and based on the parameter adjustment
- the strategy adjusts the current working parameters of the lidar, which can automatically adjust the working parameters of the lidar according to different scenarios, so that the working state of the lidar is automatically adjusted to the working state suitable for the current scene, thereby improving the working efficiency of the lidar, and can also Real-time identification of the current scene and detection of the drivable area, so as to adjust the scanning range of the lidar to match the drivable area, not only improves the obstacle detection rate, but also improves the utilization of computing resources and reduces power consumption.
- An embodiment of the present invention further provides a computer storage medium, where at least one executable instruction is stored in the storage medium, and the executable instruction causes the processor to execute the laser radar parameter adjustment method in any of the foregoing method embodiments.
- An embodiment of the present invention also provides a computer program product, the computer program product includes a computer program stored on a computer storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer, the The computer executes the laser radar parameter adjustment method in any of the above method embodiments.
- FIG. 11 shows a schematic structural diagram of a laser radar provided by an embodiment of the present invention.
- the specific embodiment of the present invention does not limit the specific implementation of the laser radar.
- the lidar may include: a transmitting device 701 , a receiving device 702 , a processor (processor) 703 , a communication interface (Communications Interface) 704 , a memory (memory) 706 , and a communication bus 708 .
- the transmitting device 701 is used for transmitting outgoing laser light to the detection area, and the receiving device 702 is used for receiving echo laser light reflected by objects in the detection area.
- the transmitting device 701 is specifically configured to scan and transmit the outgoing laser light to the detection area, so as to scan the objects in the detection area.
- the processor 703 , the communication interface 704 , and the memory 706 communicate with each other through the communication bus 708 .
- the communication interface 704 is used to communicate with network elements of other devices such as clients or other servers.
- the processor 703 is connected to the transmitting device 701 through the communication receiving 704, and is used for executing the program 710. Specifically, the laser radar parameter adjustment method in any of the above method embodiments can be executed.
- the program 710 may include program code including computer operation instructions.
- the processor 703 may be a central processing unit (CPU), or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention.
- the one or more processors included in the computing device may be the same type of processors, such as one or more CPUs; or may be different types of processors, such as one or more CPUs and one or more ASICs.
- the memory 706 is used to store the program 710 .
- Memory 706 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk memory.
- the three-dimensional environment information around the lidar is obtained, based on the three-dimensional environment information, the scene type and the drivable area where the lidar is located is identified, and the parameter adjustment strategy of the lidar is determined according to the scene type and the drivable area, and based on the parameter adjustment
- the strategy adjusts the current working parameters of the lidar, which can automatically adjust the working parameters of the lidar according to different scenarios, so that the working state of the lidar is automatically adjusted to the working state suitable for the current scene, thereby improving the working efficiency of the lidar, and can also Real-time identification of the current scene and detection of the drivable area, so as to adjust the scanning range of the lidar to match the drivable area, not only improves the obstacle detection rate, but also improves the utilization of computing resources and reduces power consumption.
- modules in the device in the embodiment can be adaptively changed and arranged in one or more devices different from the embodiment.
- the modules or units or components in the embodiments may be combined into one module or unit or component, and further they may be divided into multiple sub-modules or sub-units or sub-assemblies. All features disclosed in this specification (including accompanying claims, abstract and drawings) and any method so disclosed may be employed in any combination, unless at least some of such features and/or procedures or elements are mutually exclusive. All processes or units of equipment are combined.
- Each feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Electromagnetism (AREA)
- Aviation & Aerospace Engineering (AREA)
- Optical Radar Systems And Details Thereof (AREA)
- Traffic Control Systems (AREA)
Abstract
一种激光雷达参数调整方法、装置及激光雷达。其中,激光雷达参数调整方法包括:获取激光雷达周围三维环境信息(110);基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域(120);根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数(130)。该方法能够根据不同的场景自动调整激光雷达的工作参数。
Description
本发明实施例涉及雷达技术领域,具体涉及一种激光雷达参数调整方法、装置及激光雷达。
目前,激光雷达因为其优异的特性以及对外界环境的强适应性,在自动驾驶、辅助驾驶等领域取得了广泛的应用。
现有的激光雷达普遍保持固定的工作频率、探测角度和范围,不具备动态调整工作参数的能力,从而使得不管在什么类型的场景中工作时,激光雷达都保持相同的参数。
发明内容
本发明实施例一个目的旨在提供一种激光雷达参数调整方法、装置及激光雷达,能够根据不同的场景自动调整激光雷达的工作参数。
根据本发明实施例的一个方面,提供了一种激光雷达参数调整方法,包括:获取激光雷达周围三维环境信息;基于所述三维环境信息,识别所述激光雷达所处的场景类型以及可行驶区域;根据所述场景类型和所述可行驶区域,确定所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
在一种可选的方式中,所述三维环境信息包括三维点云数据;所述基于所述三维环境信息,识别所述激光雷达所处的场景类型以及可行驶区域,具体包括:对所述三维点云数据进行处理,生成多通道点云特征图;从所述多通道点云特征图提取高维特征信息;根据所述高维特征信息,确定所述场景类型;根据所述高维特征信息,确定所述可行驶区域。
在一种可选的方式中,所述从所述多通道点云特征图提取高维特征信息,具体包括:将所述多通道点云特征图输入第一神经网络,并获取所述第一神经网络输出的所述高维特征信息;
所述根据所述高维特征信息,确定所述场景类型,具体包括:将所述高维特征信息输入第二神经网络,并获取所述第二神经网络输出的场景类型输出值;根据预设场景类型输出值与场景类型标签的对应关系,确定所述场景类型输出值对应的所述场景类型;
所述根据所述高维特征信息,确定所述可行驶区域,具体包括:将所述高维特征信息输入第三神经网络,并获取所述第三神经网络输出的可行驶区域输出图;根据预设可行驶区域输出图与可行驶区域标签的对应关系,确定所述可行驶区域输出图对应的所述可行驶区域。
在一种可选的方式中,所述根据所述场景类型和所述可行驶区域,确定所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数,具体包括:根据所述场景类型,确定所述激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略;根据所述可行驶区域,调整所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
在一种可选的方式中,所述根据所述可行驶区域,调整所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数,具体包括:根据所述场景类型,获取所述场景类型下的标准行驶区域以及所述标准行驶区域对应的所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略;将所述标准行驶区域和所述可行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,调整所述场景参数调整策略;基于调整后的所述场景参数调整策略调整所述激光雷达的当前工作参数。
在一种可选的方式中,所述将所述标准行驶区域和所述可行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,调整所述场景参数调整策略,具体包括:分析所述可行驶区域的拓扑结构以及所述标准行驶区域的拓扑结构;计算所述可行驶区域的拓扑结构和所述标准行驶区域的拓扑结构的拓扑相关性;根据所述拓扑相关性,确定所述可行驶区域相对于所述标准行驶区域的场景差异系数;根据所述场景差异系数以及所述行驶参数调整策略,调整所述场景参数调整策略。
在一种可选的方式中,所述激光雷达的当前工作参数包括:所述激光雷达的发射光束的垂直视场角、所述激光雷达的发射光束的水平视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描频率以及激光雷达的脉冲发射频率中的一种或多种。
根据本发明实施例的另一方面,提供了一种激光雷达参数调整装置,包括:环境信息获取模块,用于获取激光雷达周围三维环境信息;识别模块,用于基于所述三维环境信息,识别所述激光雷达所处的场景类型以及可行驶区域;调整模块,用于根据所述场景类型和所述可行驶区域,确定所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
根据本发明实施例的又一方面,提供了一种激光雷达,包括:发射装置、接收装置、处理器、存储器、通信接口和通信总线,所述处理器、所述存储器和所述通信接口通过所述通信总线完成相互间的通信;所述发射装置用于发射出射激光至探测区域;所述接收装置用于接收所述探测区域内的物体反射的回波激光;所述存储器用于存放至少一可执行指令,所述可执行指令使所述处理 器执行根据如上所述的激光雷达参数调整方法的步骤。
根据本发明实施例的另一方面,提供了一种激光雷达参数调整方法,应用于包括多个激光雷达的激光雷达***,包括:
获取所述激光雷达***周围三维环境信息;
基于所述三维环境信息,识别所述激光雷达***所处的场景类型以及可行驶区域;
根据所述场景类型,确定所述激光雷达***中至少一个所述激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略;
根据所述可行驶区域,调整至少一个所述激光雷达的所述场景参数调整策略,并将调整后的场景参数调整策略确定为至少一个所述激光雷达的参数调整策略,并基于所述参数调整策略调整至少一个所述激光雷达的当前工作参数。
在一种可选的方式中,所述获取所述激光雷达***周围三维环境信息,包括:
通过所述激光雷达***中的一个激光雷达获取所述激光雷达***周围三维环境信息;其中,所述一个激光雷达为探测中远探测视场的激光雷达;
或者,通过所述激光雷达***中的多个激光雷达分别获取所述激光雷达***周围三维环境信息,其中,所述多个激光雷达对应不同的探测视场,整合所述多个激光雷达分别获取的所述三维环境信息,得到所述激光雷达***完整的三维环境信息。
在一种可选的方式中,基于所述三维环境信息,识别所述激光雷达***所处的场景类型以及可行驶区域之后,所述方法还包括:
根据所述场景类型确定与所述场景类型对应的扫描区域范围;
确定激光雷达***中各个激光雷达的扫描区域范围;
根据所述场景类型对应的扫描区域范围和所述激光雷达***中各个激光雷达的扫描区域范围确定所述激光雷达***中处于工作状态的激光雷达。
在一种可选的方式中,所述根据所述场景类型对应的扫描区域范围和所述激光雷达***中各个激光雷达的扫描区域范围确定所述激光雷达***中处于工作状态的激光雷达,包括:
当识别的所述场景类型为高速公路的场景时,选择所述激光雷达***中用于探测中远探测视场的激光雷达工作;
当识别的所述场景类型为十字路口场景时,选择激光雷达***中所有激光雷达工作。
在一种可选的方式中,所述根据所述场景类型,确定所述激光雷达***中 至少一个所述激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略,包括:
获取所述激光雷达***中的各个激光雷达的标识;
根据预设的所述场景类型、所述激光雷达***中至少一个所述激光雷达的标识和所述场景参数调整策略的对应关系,确定与所述场景类型以及所述激光雷达***中至少一个激光雷达的标识对应的所述激光雷达***中至少一个激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略。
在一种可选的方式中,所述根据所述可行驶区域,调整至少一个所述激光雷达的所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为至少一个所述激光雷达的参数调整策略,并基于所述参数调整策略调整至少一个所述激光雷达的当前工作参数,具体包括:
根据所述场景类型,获取所述场景类型下的标准行驶区域以及所述标准行驶区域对应的所述激光雷达***中至少一个所述激光雷达的水平视场角以及垂直视场角的行驶参数调整策略;
将所述标准行驶区域和所述可行驶区域进行匹配,根据匹配的结果以及所述激光雷达***中至少一个所述激光雷达的行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的场景参数调整策略;
基于所述激光雷达***中至少一个所述激光雷达的场景参数调整策略调整所述激光雷达***中至少一个所述激光雷达的当前工作参数。
在一种可选的方式中,所述将所述标准行驶区域和所述可行驶区域进行匹配,根据匹配的结果以及所述激光雷达***中至少一个所述激光雷达的行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的场景参数调整策略,包括:
分析所述可行驶区域的拓扑结构以及所述标准行驶区域的拓扑结构;
计算所述可行驶区域的拓扑结构和所述标准行驶区域的拓扑结构的拓扑相关性;
根据所述拓扑相关性,确定所述可行驶区域相对于所述标准行驶区域的场景差异系数;
根据所述场景差异系数以及所述激光雷达***中至少一个所述激光雷达的所述行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的所述场景参数调整策略。
在一种可选的方式中,所述根据所述场景差异系数以及所述激光雷达*** 中至少一个所述激光雷达的所述行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的所述场景参数调整策略,包括:
获取所述激光雷达***中至少一个所述激光雷达的所述行驶参数调整策略中的标准视场角参数;
根据所述场景差异系数,在所述标准视场角参数的基础上等比例计算新的视场角参数;
根据所述新的视场参数,调整所述激光雷达***中至少一个所述激光雷达的所述场景参数调整策略中的垂直视场角和水平视场角。
根据本发明实施例的另一方面,提供了一种激光雷达***,包括多个激光雷达,所述多个激光雷达设于激光雷达***承载装置的同一平面上,或者设于所述激光雷达***承载装置的不同平面上,所述激光雷达包括:发射装置、接收装置、处理器、存储器、通信接口和通信总线,所述处理器、所述存储器和所述通信接口通过所述通信总线完成相互间的通信;
所述发射装置用于发射出射激光至探测区域;
所述接收装置用于接收所述探测区域内的物体反射的回波激光;
所述存储器用于存放至少一可执行指令,所述可执行指令使所述处理器执行根据如上所述的激光雷达参数调整方法的步骤,对所述发射装置的工作参数进行调整。
根据本发明实施例的又一方面,提供了一种计算机存储介质,所述存储介质中存储有至少一可执行指令,所述可执行指令使处理器执行根据如上所述的激光雷达参数调整方法的步骤。
本发明实施例获取激光雷达周围三维环境信息,基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域,根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数,能够根据不同的场景自动调整激光雷达的工作参数,使激光雷达的工作状态自动调整到适合当前场景的工作状态,从而提高激光雷达的工作效率,并且,还能够实时识别当前场景和检测出可行驶区域,从而调整激光雷达的扫描范围与可行驶区域适配,不仅能够提高障碍物检测率,还能够提升计算资源利用率,降低功耗。
上述说明仅是本发明实施例技术方案的概述,为了能够更清楚了解本发明实施例的技术手段,而可依照说明书的内容予以实施,并且为了让本发明实施例的上述和其它目的、特征和优点能够更明显易懂,以下特举本发明的具体实施方式。
通过阅读下文优选实施方式的详细描述,各种其他的优点和益处对于本领域普通技术人员将变得清楚明了。附图仅用于示出优选实施方式的目的,而并 不认为是对本发明的限制。而且在整个附图中,用相同的参考符号表示相同的部件。在附图中:
图1示出了本发明实施例提供的应用场景的结构示意图;
图2示出了本发明实施例提供的一种激光雷达参数调整方法的流程示意图;
图3a-图3c示出了本发明实施例中数据处理不同阶段的点云特征图;
图4示出了本发明实施例中的神经网络训练的流程示意图;
图5a示出了本发明实施例中减小激光雷达的发射光束的垂直视场角的光路示意图;
图5b示出了本发明实施例中降低激光雷达的发射光束的光轴方向的光路示意图;
图6a和图6b示出了本发明实施例提供的可行驶区域的拓扑结构的示意图;
图7示出了本发明实施例提供的参数调整前后激光雷达的探测范围的示意图;
图8a示出了本发明另一实施例提供的应用场景的结构示意图;
图8b示出了本发明另一实施例提供的包括3个激光雷达的应用场景的示意图;
图8c示出了本发明另一实施例提供的包括5个激光雷达的应用场景的示意图;
图9示出了本发明另一实施例提供的一种激光雷达参数调整方法的流程示意图;
图10示出了本发明实施例提供的一种激光雷达参数调整装置的结构示意图;
图11示出了本发明实施例提供的一种激光雷达的结构示意图。
下面将参照附图更详细地描述本发明的示例性实施例。虽然附图中显示了本发明的示例性实施例,然而应当理解,可以以各种形式实现本发明而不应被这里阐述的实施例所限制。相反,提供这些实施例是为了能够更透彻地理解本发明,并且能够将本发明的范围完整的传达给本领域的技术人员。
图1示出了本发明实施例提供的应用场景示意图。如图1所示,该应用场景包括:激光雷达101和交通工具102。激光雷达101装设于各种需要进行周围环境探测的交通工具102上,交通工具102例如可以为车辆、船舶、飞行器等。
其中,当交通工具102为车辆时,激光雷达可以装设在车辆的车头、车尾、车顶、车侧身或其他任意可进行激光雷达固定的车身位置,可以理解的是激光 雷达也可以固定于其他与车辆进行外接的装置结构平台上。其中,可以理解的是所述外接的装置结构平台在车辆运行时与车辆可以保持同步运行。
图2示出了本发明实施例提供的一种激光雷达参数调整方法的流程示意图。该方法应用于图1中的激光雷达101。如图2所示,该方法包括:
步骤110、获取激光雷达周围三维环境信息。
其中,激光雷达可以为旋转式机械激光雷达或者固态激光雷达。可以理解的是,当激光雷达为旋转式机械激光雷达时,三维环境信息覆盖激光雷达周围360度视场范围,当激光雷达为固态激光雷达时,三维环境信息覆盖激光雷达前方一定角度范围,例如为120度。
其中,三维环境信息可以为三维点云数据。其中,可以理解的是,当激光雷达发射的激光束照射到物体表面上并被物体反射回来被接收器接收时,被接收的激光信号以点的形式被记录,便形成了三维点云数据。其中,所述三维点云数据可以包括几何位置信息、反射强度信息。其中,所述反射强度信息为激光雷达接收装置采集到的回波强度,此强度信息与目标的表面材质、粗糙度、入射角方向,以及仪器的发射能量,激光波长有关。
可以理解的是,在一些实施例中,所述三维环境信息还可以为三维点云数据与图像数据的结合数据,通过图像传感器获取彩色影像,然后将对应位置的像素的颜色信息(RGB)赋予点云中对应的点中,从而使三维点云数据具有颜色信息。其中,所述图像传感器可以集成于激光雷达内部,从而在点云数据输出前就完成这个点云数据对应颜色信息的获取;从而进行三维环境语义特征的识别;可选的,图像传感器也可以独立设置于雷达的外部,将点云数据和图像数据分别整合为三维环境信息。其中,所述图像数据可以是图像数据也可以为影像数据。
可以理解的是,激光雷达周围三维环境信息是实时获取的。
步骤120、基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域。
其中,激光雷达所处的场景类型与激光雷达所运用的领域相关,当激光雷达运用在不同的领域时,其所处的场景也不相同,因此场景类型也可以包括多种。当激光雷达运用于车辆时,场景类型包括但不限于:高速公路场景类型、十字路口场景类型、停车场场景类型等。
基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域,具体可以包括:
步骤121、对三维点云数据进行处理,生成多通道点云特征图;
其中,对三维点云数据进行处理,生成多通道点云特征图,具体包括:
将获取的三维点云数据投影到XY平面上,然后按照预设栅格大小对一定范围内的区域进行栅格化,栅格保存不同的特征值,形成多通道特征图。其中, 多通道特征图中的每一通道代表某一维度的特征值。
其中,多个通道所对应的特征值可以包括而不限于以下几种:占据率,点个数,最低点高度,最高点高度,平均高度,平均强度,距离,方位角等等。
其中,具体实施中,可以确定栅格大小0.2米,前后左右范围各80米,形成800×800大小特征图,选择多个上述特征组合作为点云多通道特征图。例如,选择占据率,点个数,平均高度和平均强度四个特征,则会构成800×800×4大小点云特征图。
步骤122、从多通道点云特征图提取高维特征信息;
其中,从多通道点云特征图提取高维特征信息,具体包括:将多通道点云特征图输入第一神经网络,所述第一神经网络通过卷积运算得到高维特征信息,其中,所述高维特征信息可以包括:边缘特征信息,形状特征信息,纹理特征信息,甚至是具有语义的特征等,例如车的轮廓信息、收费站轮廓信息,红绿灯轮廓信息,道路交通指示牌信息、路沿信息、车道线信息、人行横道信息等。其中,可以理解的是,当所述三维环境信息为三维点云数据与图像数据的结合数据时,所述从多通道点云特征图提取的高维特征信息具有颜色特征,可以进行道路指示牌、红绿灯等这种具有文字或者颜色的语义特征识别。
步骤123、根据所述高维特征信息,确定场景类型;
其中,所述根据所述高维特征信息,确定场景类型,具体包括:将所述高维特征信息输入第二神经网络,得到场景类型输出值,根据所述场景类型输出值与场景类型标签的对应关系,将所述场景类型输出值与场景类型标签匹配,确定所述场景类型输出值对应的场景类型,从而确定所述激光雷达所处的场景类型。其中,可以理解的是,场景类型标签包括与所述场景类型标签的场景类型输出值范围。可以理解的是,所述根据预设场景类型输出值与场景类型标签的对应关系,将所述场景类型输出值与场景类型标签匹配,具体包括:将所述场景类型输出值与场景类型标签对应的场景类型输出值范围进行匹配,当所述场景类型输出值落入所述场景类型标签对应的场景类型输出值范围时,则确定所述场景类型为该场景类型标签对应的场景类型。
举例来说,若将所述高维特征为弯曲的路沿、人行横道斑马线、红绿灯轮廓信息等输入第二神经网络,得到的场景输出值为9,若所述十字路口场景标签对应的场景类型输出值范围为7-10,则该场景对应的标签为十字路口场景;在举例来说,若将所述高维特征为人行横道斑马线,红绿灯轮廓信息,输入第二神经网络得到的场景类型输出值为8,则该场景也落入十字路口场景标签对应的场景输出值范围,则该场景对应的标签为十字路口场景。
可以理解的是,通过场景类型输出值与场景类型标签的对应关系来确定所述激光雷达所处的场景类型,增加了场景类型判断的准确性。
在另一种可行的实施例中,所述将所述高维特征信息输入第二神经网络得 到场景类型输出值,具体可以包括:将所述高维特征信息输入第二神经网络,将所述高维特征信息与第二神经网络模型库中的特征信息进行匹配得到,其中,可以理解的是所述第二神经网络模型库中的特征信息具有特征值及权重值,根据匹配到的特征值及权重值得到所述场景类型输出值。
步骤124、根据所述高维特征信息,确定可行驶区域。
其中,根据所述高维特征信息,确定可行驶区域,具体包括:将所述高维特征信息输入第三神经网络,得到可行驶区域输出图,根据预设可行驶区域输出图与可行驶区域标签的对应关系,将可行驶区域输出图与可行驶区域标签进行匹配,确定所述可行驶区域输出图对应的可行驶区域,从而确定可行驶区域。
其中,可以理解的是,将所述高维特征信息输入第三神经网络,得到可行驶区域输出图,可以包括:将所述高维特征信息输入第三神经网络,第三神经网络进行可行驶区域特征提取,其中,可行驶区域特征例如可以包括路沿信息、车道线信息等;将包含所述可行驶区域特征的区域提取出来,生成可行驶区域输出图。
其中,可以理解的是,不同场景下的可行驶区域标签可以不同,举例来说,高速公路的可行驶区域标签可以为包含有车道线信息的标签,十字路口的可行驶区域标签可以包含路沿信息和车道线信息,停车场的可行驶区域标签可以包括停车的车位线信息等。
将可行驶区域输出图与可行驶区域标签进行匹配,确定可行驶区域,可以包括:将可行驶区域输出图与可行驶区域标签进行匹配,确定所述可行驶区域输出图所在的场景,根据所述场景确定可行驶区域。
其中,所述将可行驶区域输出图与可行驶区域标签进行匹配,确定可行驶区域之后,所述方法还包括:获取可行驶区域中的障碍物信息,根据所述障碍物信息确定可行驶区域的复杂度;若所述可行驶区域的复杂度超过预设阈值,则重新获取可行驶区域。
其中,可以理解的是,其中,第一神经网络和第二神经网络均由卷积层和池化层组成,第三神经网络由反卷积层组成。高维特征信息为中间变量,由第一神经网络通过卷积运算得到,并分别输入到第二神经网络和第三神经网络中。例如,如图3a-3c所示,为不同处理阶段的点云特征图,其中图3a为数据处理初期的点云特征图(原始的三维点云数据),图3b为数据处理中期的点云特征图(经过第一神经网络后的输出),图3c为数据处理完成的点云特征图(经过第二神经网络后的输出)。
其中,由于神经网络的基础运算为卷积运算,其由一系列乘加操作组成,为了引入非线性变换,神经网络中会加入非线性激活函数。常用的非线性激活函数有ReLU、Sigmoid等。
可选的,在一些可选的实施例中,在步骤120之前,该方法还包括:获取 训练点云数据,训练点云数据中包含原始点云数据和与该原始点云数据对应的场景类型标签和可行驶区域标签;根据原始点云数据和与该原始点云数据对应的场景类型标签和可行驶区域标签,对第一神经网络、第二神经网络、第三神经网络进行训练。
其中,原始点云数据可以由车载激光雷达得到,与该原始点云数据对应的场景类型标签和可行驶区域标签可以由人工标注得到。其中,原始点云数据形成的训练集中的原始点云数据量越大,训练得到的神经网络就越准确。
其中,如图4所示,对第一神经网络、第二神经网络、第三神经网络进行训练,具体可以包括:分别获取第一神经网络、第二神经网络、第三神经网络的初始化参数;将训练点云数据中的原始点云数据生成训练点云特征图;将训练点云特征图输入第一神经网络、第二神经网络、第三神经网络,并获取第二神经网络输出的训练场景类型、第三神经网络输出的训练感兴趣区域;将训练场景类型与场景类型标签比较,通过第二损失函数计算场景类型损失值;将训练感兴趣区域与可行驶区域标签比较,通过第三损失函数计算可行驶区域损失值;基于场景类型损失值和可行驶区域损失值,通过反向传播算法对第一神经网络、第二神经网络、第三神经网络进行参数更新;当场景类型损失值和可行驶区域损失值不再变化时停止训练,保存最终的神经网络参数值。
步骤130、根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数。
其中,在本实施例中,激光雷达的当前工作参数包括但不限于:激光雷达的发射光束的垂直视场角、激光雷达的发射光束的水平视场角、激光雷达的发射光束的光轴方向、激光雷达的扫描密度、激光雷达的扫描频率以及激光雷达的脉冲发射功率中的一种或多种。
其中,步骤130具体可以包括:
步骤131、根据所述场景类型,确定激光雷达的水平视场角、激光雷达的垂直视场角、激光雷达的发射光束的光轴方向、激光雷达的扫描密度、激光雷达的扫描频率以及激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略。
其中,场景参数调整策略以及场景类型与场景参数调整策略的对应关系由用户根据场景类型预先设置。
例如,当装载有激光雷达的车辆行驶在高速公路场景中,需要激光雷达将扫描范围集中在前后方道路区域内,减少对车辆左右方的探测,忽略或者减少对可行驶区域外的探测。缩小探测范围不仅能够节省计算力,还能提高重点区域内的障碍物检测率。因此,在一些可选的实施例中,当检测到为高速公路场景类型时,则减小激光雷达的垂直视场角、降低激光雷达的发射光束的脉冲发射频率和提高激光雷达的扫描频率。其中,减小发射光束的垂直视场角,如图5a,例如可以压缩20%垂直视场角,或者,将激光雷达的垂直视场角从+15°~ -25°调整到+12°~-20°,能够将更多线束集中在水平方向,增大激光雷达的前后探测距离,从而探测到远距离目标。其中,降低脉冲发射频率,例如将脉冲发射频率从每秒发射5万个调整到每秒发射1万个,能够增加脉冲激光发射器的充能时间,使得激光能量增增加,从而探测更远的距离。其中,提高扫描频率,例如将扫描频率从10Hz调整到15Hz,能够更快地探测到移动物体的变化,以提高自动驾驶的安全性。
例如,当装载有激光雷达的车辆位于十字路口场景中时,通过十字路口时,周围环境中的物体移动变化快,此时需要更快地探测到移动物体的变化。当检测到为十字路口场景类型时,则需要减少前后方的探测距离,增加车周边近距离的探测,以满足更快的探测到移动物体的变化。因此,在一些可选的实施例中,当检测到为十字路口场景时,则提高激光雷达的脉冲发射频率、降低激光雷达的发射光束的光轴方向以及提高激光雷达的扫描频率。其中,提高激光雷达的脉冲发射频率,例如将脉冲发射频率从每秒发射1万个调整到每秒发射5万个,从而能够更快的获取周围移动物体的变化。其中,降低发射光束的光轴方向,如图5b,例如可以发射光束的光轴方向下移20°,在一具体应用实例中将发射光束的光轴方向从0°(也即水平出射)向下偏转20°,发射光束的探测范围将更靠近地面,能够减少激光雷达的探测距离,从而集中探测近距离目标。其中,提高激光雷达的扫描频率,例如可以将扫描频率从10Hz调整到15Hz,能够更快地探测到移动物体的变化。
例如,当装载有激光雷达的车辆进入停车场等静态场景或者停车等待时,在周围固定不变的情况下,可以减少激光雷达的功率消耗。因此,在一些可选的实施例中,当检测到为停车场场景类型时,则降低激光雷达的脉冲发射频率、降低激光雷达的发射光束的光轴方向以及降低激光雷达的扫描频率。其中,降低激光雷达的扫描频率,例如可以将扫描频率从15Hz调整到10Hz,能够减少激光雷达的功率消耗。
步骤132、根据所述可行驶区域,调整所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
其中,步骤132具体可以包括:
步骤1321、根据所述场景类型,获取所述场景类型下的标准行驶区域以及所述标准行驶区域对应的所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略。
其中,预先设置场景类型与标准行驶区域的对应关系,则能够根据当前场景类型,在场景类型与标准行驶区域的对应关系中查找当前场景类型下的标准行驶区域。例如,预先设定高速公路场景类型对应的标准行驶区域为某特定的直线道路区域,则当确定场景类型为高速公路场景类型,查找到对应的标准行驶区域为某特定的直线道路区域。同时,场景类型、标准行驶区域、行驶参数 调整策略存在对应关系,在理想情况下,在某个场景类型的可行驶区域应该为标准行驶区域,而根据该标准行驶区域应该设置预设标准视场角参数,然而,实际确定的可行驶区域不一定与标准行驶区域完全吻合,因此,还预先设置标准行驶区域与所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略的对应关系,则能够根据获取的标准行驶区域,在标准行驶区域与所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略的对应关系中查找获取的标准行驶区域对应的所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略。
步骤1322、将所述可行驶区域与标准行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,对所述场景参数调整策略进行再调整。
其中,在一些可选的实施例中,将所述标准行驶区域和所述可行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,调整所述场景参数调整策略,具体包括:分析可行驶区域的拓扑结构以及标准行驶区域的拓扑结构,并计算可行驶区域的拓扑结构与标准行驶区域的拓扑结构的拓扑相关性;根据所述拓扑相关性,确定可行驶区域相对于标准行驶区域的场景差异系数;其中,所述场景差异系数包括的平移差异系数、旋转差异系数和缩放差异系数等;根据所述场景差异系数以及所述行驶参数调整策略,调整所述场景参数调整策略。其中,在一些实施例中,根据所述场景差异系数以及所述行驶参数调整策略,调整所述场景参数调整策略,具体包括:获取行驶参数调整策略中的标准视场角参数(包括标准垂直视场角参数及标准水平视场角参数);根据场景差异系数,在标准视场角参数的基础上等比例计算新的视场角参数(包括新的垂直视场角及新的水平视场角参数);根据新的视场参数,调整场景参数调整策略中的垂直视场角和水平视场角。例如,如图6a所示,假设可行驶区域的拓扑结构和标准行驶区域的拓扑结构的中心重合、边平行、宽度相同、长度之比为5:3,则确定可行驶区域的拓扑结构和标准行驶区域的拓扑结构的平移差异系数为0、旋转差异系数为0、缩放差异系数为(5/3)*1。又例如,如图6b所示,假设可行驶区域的拓扑结构和标准行驶区域的拓扑结构的中心重合、边平行、宽度之比为6:4、长度之比为6:4,则确定可行驶区域的拓扑结构和标准行驶区域的拓扑结构的平移差异系数为0、旋转差异系数为0、缩放差异系数为(6/4)*(6/4)。可选地,在一些其他实施例中,当可行驶区域的拓扑结构和标准行驶区域的拓扑结构的中心不重合或者边不平行,则可以保持标准行驶区域的拓扑结构不变,对可行驶区域的拓扑结构进行旋转、平移,直至可行驶区域的拓扑结构和标准行驶区域的拓扑结构的中心重合且边平行,并记录平移差异系数和旋转差异系数,然后再计算可行驶区域的拓扑结构和标准行驶区域的拓扑结构的长度和宽度的比例关系,确定缩放差异系数。
其中,当计算得到平移差异系数、旋转差异系数和缩放差异系数后,可以根据平移差异系数、旋转差异系数和缩放差异系数,计算新的视场角参数。例如,当计算得到平移差异系数、旋转差异系数和缩放差异系数,则可以根据平 移差异系数、旋转差异系数、缩放差异系数、以及相似图形的边的比例关系,计算得到视场角的放大倍数,从而计算得到新的视场角参数,以使得当调整到新的视场角参数时,激光雷达的探测范围能够完全覆盖标准行驶区域。又例如,当计算得到平移差异系数、旋转差异系数和缩放差异系数,可以根据旋转差异系数调整激光雷达的发射光束的光轴方向,以使得调整后的激光雷达的发射光束的光轴方向平行标准行驶区域的拓扑结构的中心轴,然后再根据平移差异系数、缩放差异系数、以及相似图形的边的比例关系,计算得到视场角的放大倍数,从而计算得到新的视场角参数,以使得当调整到新的视场角参数时,激光雷达的探测范围能够完全覆盖标准行驶区域。
其中,可以理解的是,雷达调整状态后,新的检测范围将与可行驶区域按照上述方法再次计算差异系数,在此基础进一步调整视场角参数(垂直视场角及水平视场角参数)。
步骤1323、基于调整后的所述场景参数调整策略调整所述激光雷达的当前工作参数。
需要说明的是,参数调整幅度或者预设的调整后的参数值不能超过激光雷达可执行的范围。在一些可选的实施例中,当参数调整幅度或者预设的调整范围超过激光雷达可执行的范围时,提示重新调整监控区域。
例如,确定与高速公路场景类型匹配的场景参数调整策略包括:将激光雷达的发射光束的垂直视场角从+45°压缩到40°、将激光雷达的发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将激光雷达的扫描频率从10Hz调整到15Hz,以及与高速公路场景类型匹配的行驶参数调整策略包括:水平视场角为120°、垂直视场角为35°,假设可行驶区域的拓扑结构与标准行驶区域的拓扑结构的拓扑相关性为中心重合、边平行、宽度相同、长度之比为3:5,即平移差异系数0、旋转差异系数为0、缩放差异系数为(3/5)*1,则计算得到新的水平视场角为(3/5)*120°=72°、新的垂直视场角为35°(如图7所示),最后得到调整后的场景参数调整策略包括:将激光雷达的发射光束的水平视场调整为72°、将激光雷达的发射光束的垂直视场角调整为35°、将激光雷达的发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将激光雷达的扫描频率从10Hz调整到15Hz。
例如,确定与十字路口场景类型匹配的场景参数调整策略包括:将脉冲发射频率从每秒发射1万个调整到每秒发射5万个、将发射光束的光轴方向下移20°、将扫描频率从10Hz调整到15Hz,以及与十字路口场景类型匹配的行驶参数调整策略包括:水平视场角为120°、垂直视场角为90°,假设可行驶区域的拓扑结构与标准行驶区域的拓扑结构的拓扑相关性为中心重合、边平行、宽度之比为6:4、长度之比为6:4,即平移差异系数为0、旋转差异系数为0、缩放差异系数为(6/4)*(6/4),则计算得到新的水平视场角为(6/4)*120°=180°、新的垂直视场角为(6/4)*90°=135°,最后得到调整后的场景参数调整策略包 括:将激光雷达的发射光束的水平视场角调整为180°,将激光雷达的发射光束的垂直视场角调整为135°,将脉冲发射频率从每秒发射1万个调整到每秒发射5万个,将发射光束的光轴方向下移20°,将扫描频率从10Hz调整到15Hz。
例如,确定与停车场场景类型匹配的场景参数调整策略包括:将激光雷达的发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将激光雷达的发射光束的光轴方向降低20°、将扫描频率从15Hz降低为10Hz,以及与停车场场景类型匹配的行驶参数调整策略包括:水平视场角为120°、垂直视场角为60°,假设可行驶区域的拓扑结构与标准行驶区域的拓扑结构的拓扑相关性为中心重合、边平行、宽度之比为1:2、长度之比为1:2,即平移差异系数0、旋转差异系数为0、缩放差异系数为(1/2)*(1/2),则计算得到新的水平视场角为(1/2)*120°=60°、新的垂直视场角为(1/2)*60°=30°,最后得到调整后的场景参数调整策略包括:将激光雷达的发射光束的水平视场调整为60°,将激光雷达的发射光束的垂直视场角调整为30°,将激光雷达的发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将激光雷达的发射光束的光轴方向降低20°、将扫描频率从15Hz降低为10Hz。
本发明实施例通过获取激光雷达周围三维环境信息,基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域,根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数,能够根据不同的场景自动调整激光雷达的工作参数,使激光雷达的工作状态自动调整到适合当前场景的工作状态,从而提高激光雷达的工作效率,并且,还能够实时识别当前场景和检测出可行驶区域,从而调整激光雷达的扫描范围与可行驶区域适配,不仅能够提高障碍物检测率,还能够提升计算资源利用率,降低功耗。
图8a示出了本发明另一实施例提供的应用场景的结构示意图。如图8a所示,该应用场景包括:激光雷达***103和交通工具102。激光雷达***103装设于各种需要进行周围环境探测的交通工具102上,交通工具102可以为车辆、船舶、飞行器等。
其中,激光雷达***103包括多个激光雷达。多个激光雷达可以设于同一平面上,也可以设于不同平面上。当交通工具102为车辆时,激光雷达可以装设在车辆的车头、车尾、车顶中的一处或多处。例如,如图8b所示,激光雷达***103包括三个激光雷达,三个激光雷达1031、1032、1033均设置在车辆的车顶,并且,激光雷达1031位于车顶的中心,激光雷达1032、1033位于车顶的相对两侧,激光雷达1031可以用于获取车辆四周且离车辆较远的环境信息,激光雷达1032、1033分别用于获取车辆左右两侧且离车辆较近的环境信息;又例如,如图8c所示,激光雷达***103包括五个激光雷达,激光雷达1034设置在车辆的车顶、并位于车顶的中心,激光雷达1035、1036设置在车辆的左右两侧、并位于后视镜周围,激光雷达1037设置车辆的车头,激光雷达1038设置车辆的车尾。
图9示出了本发明另一实施例提供的一种激光雷达参数调整方法的流程示意图。该方法应用于图8a至图8c中的激光雷达***103。如图9所示,该方法包括:
步骤210、获取激光雷达***周围三维环境信息。
其中,步骤210的具体实施方式与上述实施例中的步骤110大致相同。在本实施例中,通过激光雷达***中的一个激光雷达获取激光雷达***周围三维环境信息,也可以通过激光雷达***中的多个激光雷达获取激光雷达***周围三维环境信息。例如,如图8b所示,当通过激光雷达***中的一个激光雷达获取激光雷达***周围三维环境信息时,具体可以为:通过激光雷达***103中的激光雷达1031获取激光雷达***103周围三维环境信息;当通过激光雷达***中的多个激光雷达获取激光雷达***103周围三维环境信息时,具体也可以为:通过激光雷达***103中的激光雷达1031、1032、1033分别获取部分三维环境信息(如激光雷达1031获取前侧三维环境信息、激光雷达1032获取右侧三维环境信息、激光雷达1033获取左侧三维环境信息),并整合获取的三维环境信息,从而得到完整的三维环境信息。
步骤220、基于三维环境信息,识别激光雷达***所处的场景类型以及可行驶区域。
其中,步骤220的具体实施方式与上述实施例中的步骤120大致相同,此处不再赘述。
步骤230、根据场景类型和可行驶区域,确定激光雷达***中各个激光雷达的参数调整策略,并基于各个激光雷达的参数调整策略分别调整各个激光雷达的当前工作参数。
具体地,步骤230包括:
步骤231、根据场景类型,确定激光雷达***中各个激光雷达的水平视场角、垂直视场角、发射光束的光轴方向、扫描密度、扫描频率以及脉冲发射功率中的一种或多种的场景参数调整策略。
其中,场景参数调整策略以及场景类型与场景参数调整策略的对应关系由用户根据场景类型预先设置。激光雷达***中的不同激光雷达的场景参数调整策略可以不同。例如,场景类型与不同激光雷达的场景参数调整策略的对应关系也可以由用户根据场景类型以及不同激光雷达的位置预先设置。在本实施例中,根据场景类型,确定激光雷达***中各个激光雷达的水平视场角、垂直视场角、发射光束的光轴方向、扫描密度、扫描频率以及脉冲发射功率中的一种或多种的场景参数调整策略,具体可以为:获取激光雷达***中的激光雷达的标识,根据预设的场景类型、激光雷达的标识和场景参数调整策略的对应关系,确定与识别的场景类型以及获取的激光雷达的标识对应的场景参数调整策略。
例如,如图8b所示,假设激光雷达1031、1032、1033的标识分别为a、b、 c,则确定与识别的场景类型以及标识a、b、c对应的场景参数调整策略,当检测到当前的场景类型为高速公路场景类型时,确定a对应的场景参数调整策略为:减小激光雷达的垂直视场角、降低激光雷达的发射光束的脉冲发射频率和提高激光雷达的扫描频率,确定b、c对应的场景参数调整策略为:减小激光雷达的水平视场角、降低激光雷达的发射光束的脉冲发射频率和提高激光雷达的扫描频率,从而增大车顶的激光雷达的前后探测距离,压缩车顶两侧的激光雷达的左右探测距离,并且更快的检测到移动物体的变化;当检测到当前的场景类型为十字路口场景类型时,确定a对应的场景参数调整策略为:增大激光雷达的垂直视场角、提高激光雷达的脉冲发射频率、降低激光雷达的发射光束的光轴方向以及提高激光雷达的扫描频率,确定b、c对应的场景参数调整策略为:减小激光雷达的水平视场角、提高激光雷达的脉冲发射频率以及提高激光雷达的扫描频率,从而车顶的激光雷达的前后探测距离,使车顶上的激光雷达的发射光束的探测范围更靠近地面,压缩车顶两侧的激光雷达的左右探测距离,并且使所有激光雷达更快地探测到移动物体的变化;当检测到当前的场景类型为停车场场景类型时,确定a对应的场景参数调整策略为:减小激光雷达的垂直视场角、降低激光雷达的脉冲发射频率以及降低激光雷达的扫描频率,确定b、c对应的场景参数调整策略为:减小激光雷达的水平视场角、降低激光雷达的脉冲发射频率以及降低激光雷达的扫描频率,从而减小激光雷达的探测范围,并且减少激光雷达的功率消耗。
在一些可选的实施例中,可以根据所述各个激光雷达的扫描区域范围确定所述各个激光雷达的参数调整策略。
其中,所述将标准行驶区域和可行驶区域进行匹配,根据匹配的结果以及激光雷达***中各个激光雷达的行驶参数调整策略,分别调整激光雷达***中各个激光雷达的场景参数调整策略之前,所述方法还包括:获取所述激光雷达***中各个激光雷达的扫描区域范围;根据所述各个激光雷达的扫描区域范围确定所述各个激光雷达的参数调整策略。
其中,如图8b所示的激光雷达***,中心激光雷达1031用于探测中远距离视场的探测,激光雷达1033和激光雷达1032分别用于近距离视场探测和左右视场探测;
其中,可以理解的是,所述激光雷达1031的中轴线垂直于水平面;其中,激光雷达1031的中轴线为通过激光雷达1031的底面中心的底面法线。激光雷达1031位于所述中心激光雷达的左侧,激光雷达1033的中轴线与激光雷达1031的中轴线之间的夹角度小于零度;激光雷达1032位于激光雷达的右侧,激光雷达1032的中轴线与激光雷达1031的中轴线之间的夹角大于零度;其中,激光雷达1032和激光雷达1033的中轴线为通过所述激光雷达的底面中心的底面法线。
其中,可以理解的是,所述激光雷达***中各个激光雷达可以在同一平面 也可以在不同平面。
其中,可以理解的是,当根据识别的场景需要加强中远视场的探测需求的时候,调整激光雷达1031扫描频率、扫描密度、脉冲发射功率中的一种或多种。当根据识别的场景需要加强近场车左右的视场的探测需求的时候,可以调整左激光雷达1033和右激光雷达1032的发射光束的光轴方向、扫描密度、扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种。
其中,在另一些可选的实施例中,激光雷达***也可以根据识别的场景选择部分激光雷达工作或者全部激光雷达工作;举例来说,如图8b,当识别为高速公路场景(即更需要进行中远视场的探测),可以选择激光雷达1031工作,根据场景及可行驶区域确定激光雷达1031的调整参数。当识别为十字路口场景时(即需要增强车身周边物体的探测),则可以同时开启***中所有激光雷达的工作,调整***中(如图8b)激光雷达1033和激光雷达1032的工作参数。
其中,可以理解的是,如图8c,所述激光雷达***103也可以包括五个激光雷达,其中,中心激光雷达1031用于探测中远距离视场的探测,激光雷达1033和激光雷达1032分别用于左右近距离视场探测;1034和1035为近距离前后视场探测。
其中,当激光雷达***包括5个激光雷达的时候,激光雷达***根据所述各个激光雷达的扫描区域范围确定所述各个激光雷达的参数调整策略。具体,举例来说,如图8c,当识别为高速公路场景(即更需要进行中远视场的探测),可以选择激光雷达1031工作,根据场景及可行驶区域确定激光雷达1031的调整参数。当识别为十字路口场景时(即需要增强车身周边物体的探测),则可以同时开启***中所有激光雷达的工作,调整***中(如图8b)激光雷达1033、激光雷达1032、激光雷达1034和激光雷达1035的工作参数。
步骤232、根据可行驶区域,调整各个激光雷达的场景参数调整策略,并将调整后的各个激光雷达场景参数调整策略确定为各个激光雷达的参数调整策略,并基于各个激光雷达的参数调整策略分别调整各个激光雷达的当前工作参数。
其中,步骤232具体可以包括:
步骤2321、根据场景类型,获取场景类型下的标准行驶区域以及标准行驶区域对应的激光雷达***中各个激光雷达的水平视场角以及垂直视场角的行驶参数调整策略。
其中,预先设置场景类型与标准行驶区域的对应关系,则能够根据当前场景类型,在场景类型与标准行驶区域的对应关系中查找当前场景类型下的标准行驶区域。例如,预先设定高速公路场景类型对应的标准行驶区域为某特定的直线道路区域,则当确定场景类型为高速公路场景类型,查找到对应的标准行驶区域为某特定的直线道路区域。同时,场景类型、标准行驶区域、行驶参数调整策略存在对应关系,在理想情况下,在某个场景类型的可行驶区域应该为 标准行驶区域,根据该标准行驶区域可以设置预设标准视场角参数,然而,实际确定的可行驶区域不一定与标准行驶区域完全吻合,因此,还需要预先设置标准行驶区域与激光雷达***中各个激光雷达的水平视场角以及垂直视场角的行驶参数调整策略的对应关系,则能够根据获取的标准行驶区域,在标准行驶区域与激光雷达***中各个激光雷达的水平视场角以及垂直视场角的行驶参数调整策略的对应关系中查找获取的标准行驶区域对应的各个激光雷达的水平视场角以及垂直视场角的行驶参数调整策略。
步骤2322、将标准行驶区域和可行驶区域进行匹配,根据匹配的结果以及激光雷达***中各个激光雷达的行驶参数调整策略,分别调整激光雷达***中各个激光雷达的场景参数调整策略。
其中,在一些可选的实施例中,将标准行驶区域和可行驶区域进行匹配,根据匹配的结果以及激光雷达***中各个激光雷达的行驶参数调整策略,分别调整激光雷达***中各个激光雷达的场景参数调整策略,具体包括:分析可行驶区域的拓扑结构以及标准行驶区域的拓扑结构,并计算可行驶区域的拓扑结构与标准行驶区域的拓扑结构的拓扑相关性;根据拓扑相关性,确定可行驶区域相对于标准行驶区域的场景差异系数,其中,场景差异系数包括的平移差异系数、旋转差异系数和缩放差异系数等;根据场景差异系数以及激光雷达***中各个激光雷达的行驶参数调整策略,分别调整激光雷达***中各个激光雷达的场景参数调整策略。其中,在一些实施例中,根据场景差异系数以及激光雷达***中各个激光雷达的行驶参数调整策略,分别调整激光雷达***中各个激光雷达的场景参数调整策略,具体包括:获取某个激光雷达的行驶参数调整策略中的标准视场角参数(包括标准垂直视场角参数及标准水平视场角参数);根据场景差异系数,在标准视场角参数的基础上等比例计算新的视场角参数(包括新的垂直视场角及新的水平视场角参数);根据新的视场参数,调整该激光雷达的场景参数调整策略中的垂直视场角和水平视场角;并按照相同的方式调整激光雷达***中各个激光雷达的场景参数调整策略。
步骤2323、基于激光雷达***中各个激光雷达的场景参数调整策略调整激光雷达***中各个激光雷达的当前工作参数。
例如,如图8b所示,假设激光雷达1031、1032、1033的标识分别为a、b、c,当检测到当前的场景类型为高速公路场景类型时,确定a对应的场景参数调整策略为:将发射光束的垂直视场角从+90°压缩到80°、将发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将扫描频率从10Hz调整到15Hz,确定b、c对应的场景参数调整策略为:将发射光束的水平视场角从+180°压缩到120°、将发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将扫描频率从10Hz调整到15Hz;确定a对应的行驶参数调整策略为:水平视场角为120°、垂直视场角为60°,确定b、c对应的行驶参数调整策略为:水平视场角为100°、垂直视场角为35°;假设可行驶区域的拓扑结构与标准行驶区域的拓扑结构的拓扑相关性为中心重合、边平行、宽度相同、长度 之比为3:5,即平移差异系数0、旋转差异系数为0、缩放差异系数为(3/5)*1,计算得到a对应的新的水平视场角为(3/5)*120°=72°、新的垂直视场角为60°,b、c对应的新的水平视场角为(3/5)*100°=60°、新的垂直视场角为35°;最后得到a对应的调整后的场景参数调整策略为:将发射光束的水平视场角调整为72°、将发射光束的垂直视场角调整为60°、将发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将扫描频率从10Hz调整到15Hz,b、c对应的调整后的场景参数调整策略为:将发射光束的水平视场角调整为60°、将发射光束的垂直视场角调整为35°、将发射光束的脉冲发射频率从每秒发射5万个调整到每秒发射1万个、将扫描频率从10Hz调整到15Hz。
本发明实施例通过获取激光雷达***周围三维环境信息,基于三维环境信息,识别激光雷达***所处的场景类型以及可行驶区域,根据场景类型和可行驶区域,确定激光雷达***中各个激光雷达的参数调整策略,并基于各个激光雷达的参数调整策略分别调整各个激光雷达的当前工作参数,能够根据不同的场景自动调整激光雷达***中各个激光雷达的工作参数,使激光雷达***的工作状态自动调整到适合当前场景的工作状态,从而提高工作效率,并且,还能够实时识别当前场景和检测出可行驶区域,从而使激光雷达***中的各个激光雷达的扫描范围与可行驶区域适配,不仅能够提高障碍物检测率,还能够提升计算资源利用率,降低功耗。
图10示出了本发明实施例提供的一种激光雷达参数调整装置的结构示意图。该装置可以应用于上述实施例中的激光雷达或激光雷达***。如图10所示,该装置600包括:环境信息获取模块610、识别模块620和调整模块630。
其中,环境信息获取模块610用于获取激光雷达周围三维环境信息;
识别模块620用于基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域;调整模块630用于根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数。
在一种可选的方式中,所述三维环境信息包括三维点云数据;识别模块620具体用于:对所述三维点云数据进行处理,生成多通道点云特征图;从所述多通道点云特征图提取高维特征信息;根据所述高维特征信息,确定所述场景类型;根据所述高维特征信息,确定所述可行驶区域。
在一种可选的方式中,识别模块620具体用于:将所述多通道点云特征图输入第一神经网络,并获取所述第一神经网络输出的所述高维特征信息;将所述高维特征信息输入第二神经网络,并获取所述第二神经网络输出的场景类型输出值;根据预设场景类型输出值与场景类型标签的对应关系,确定所述场景类型输出值对应的所述场景类型;将所述高维特征信息输入第三神经网络,并获取所述第三神经网络输出的可行驶区域输出图;根据预设可行驶区域输出图与可行驶区域标签的对应关系,确定所述可行驶区域输出图对应的所述可行驶区域。
在一种可选的方式中,调整模块630具体用于:根据所述场景类型,确定所述激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略;根据所述可行驶区域,调整所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
在一种可选的方式中,调整模块630具体还用于:根据所述场景类型,获取所述场景类型下的标准行驶区域以及所述标准行驶区域对应的所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略;将所述标准行驶区域和所述可行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,调整所述场景参数调整策略;基于调整后的所述场景参数调整策略调整所述激光雷达的当前工作参数。
在一种可选的方式中,调整模块630具体还用于:分析所述可行驶区域的拓扑结构以及所述标准行驶区域的拓扑结构;计算所述可行驶区域的拓扑结构和所述标准行驶区域的拓扑结构的拓扑相关性;根据所述拓扑相关性,确定所述可行驶区域相对于所述标准行驶区域的场景差异系数;根据所述场景差异系数以及所述行驶参数调整策略,调整所述场景参数调整策略。
在一种可选的方式中,所述激光雷达的当前工作参数包括:所述激光雷达的发射光束的垂直视场角、所述激光雷达的发射光束的水平视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描频率以及激光雷达的脉冲发射频率中的一种或多种。
需要说明的是,本发明实施例提供的激光雷达参数调整装置是能够执行上述激光雷达参数调整方法的装置,则上述基于激光雷达参数调整方法的所有实施例均适用于该装置,且均能达到相同或相似的有益效果。
本实施例通过获取激光雷达周围三维环境信息,基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域,根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数,能够根据不同的场景自动调整激光雷达的工作参数,使激光雷达的工作状态自动调整到适合当前场景的工作状态,从而提高激光雷达的工作效率,并且,还能够实时识别当前场景和检测出可行驶区域,从而调整激光雷达的扫描范围与可行驶区域适配,不仅能够提高障碍物检测率,还能够提升计算资源利用率,降低功耗。
本发明实施例还提供了一种计算机存储介质,所述存储介质中存储有至少一可执行指令,所述可执行指令使处理器执行上述任意方法实施例中的激光雷达参数调整方法。
本发明实施例还提供了一种计算机程序产品,所述计算机程序产品包括存 储在计算机存储介质上的计算机程序,所述计算机程序包括程序指令,当所述程序指令被计算机执行时,使所述计算机执行上述任意方法实施例中的激光雷达参数调整方法。
图11示出了本发明实施例提供的一种激光雷达的结构示意图,本发明具体实施例并不对激光雷达的具体实现做限定。
如图11所示,该激光雷达可以包括:发射装置701、接收装置702、处理器(processor)703、通信接口(Communications Interface)704、存储器(memory)706、以及通信总线708。
其中,发射装置701用于发射出射激光至探测区域,接收装置702用于接收探测区域内的物体反射的回波激光。其中,发射装置701具体用于将出射激光扫描发射至探测区域,以扫描探测区域内的物体。
其中:处理器703、通信接口704、以及存储器706通过通信总线708完成相互间的通信。通信接口704,用于与其它设备比如客户端或其它服务器等的网元通信。处理器703,通过通信接收704与发射装置701连接,用于执行程序710,具体可以执行上述任意方法实施例中的激光雷达参数调整方法。
具体地,程序710可以包括程序代码,该程序代码包括计算机操作指令。
处理器703可能是中央处理器CPU,或者是特定集成电路ASIC(Application Specific Integrated Circuit),或者是被配置成实施本发明实施例的一个或多个集成电路。计算设备包括的一个或多个处理器,可以是同一类型的处理器,如一个或多个CPU;也可以是不同类型的处理器,如一个或多个CPU以及一个或多个ASIC。
存储器706,用于存放程序710。存储器706可能包含高速RAM存储器,也可能还包括非易失性存储器(non-volatile memory),例如至少一个磁盘存储器。
本实施例通过获取激光雷达周围三维环境信息,基于三维环境信息,识别激光雷达所处的场景类型以及可行驶区域,根据场景类型和可行驶区域,确定激光雷达的参数调整策略,并基于参数调整策略调整激光雷达的当前工作参数,能够根据不同的场景自动调整激光雷达的工作参数,使激光雷达的工作状态自动调整到适合当前场景的工作状态,从而提高激光雷达的工作效率,并且,还能够实时识别当前场景和检测出可行驶区域,从而调整激光雷达的扫描范围与可行驶区域适配,不仅能够提高障碍物检测率,还能够提升计算资源利用率,降低功耗。
在此提供的算法或显示不与任何特定计算机、虚拟***或者其它设备固有相关。各种通用***也可以与基于在此的示教一起使用。根据上面的描述,构造这类***所要求的结构是显而易见的。此外,本发明实施例也不针对任何特定编程语言。应当明白,可以利用各种编程语言实现在此描述的本发明的内容, 并且上面对特定语言所做的描述是为了披露本发明的最佳实施方式。
在此处所提供的说明书中,说明了大量具体细节。然而,能够理解,本发明的实施例可以在没有这些具体细节的情况下实践。在一些实例中,并未详细示出公知的方法、结构和技术,以便不模糊对本说明书的理解。
类似地,应当理解,为了精简本发明并帮助理解各个发明方面中的一个或多个,在上面对本发明的示例性实施例的描述中,本发明实施例的各个特征有时被一起分组到单个实施例、图、或者对其的描述中。然而,并不应将该公开的方法解释成反映如下意图:即所要求保护的本发明要求比在每个权利要求中所明确记载的特征更多的特征。更确切地说,如下面的权利要求书所反映的那样,发明方面在于少于前面公开的单个实施例的所有特征。因此,遵循具体实施方式的权利要求书由此明确地并入该具体实施方式,其中每个权利要求本身都作为本发明的单独实施例。
本领域那些技术人员可以理解,可以对实施例中的设备中的模块进行自适应性地改变并且把它们设置在与该实施例不同的一个或多个设备中。可以把实施例中的模块或单元或组件组合成一个模块或单元或组件,以及此外可以把它们分成多个子模块或子单元或子组件。除了这样的特征和/或过程或者单元中的至少一些是相互排斥之外,可以采用任何组合对本说明书(包括伴随的权利要求、摘要和附图)中公开的所有特征以及如此公开的任何方法或者设备的所有过程或单元进行组合。除非另外明确陈述,本说明书(包括伴随的权利要求、摘要和附图)中公开的每个特征可以由提供相同、等同或相似目的的替代特征来代替。
此外,本领域的技术人员能够理解,尽管在此的一些实施例包括其它实施例中所包括的某些特征而不是其它特征,但是不同实施例的特征的组合意味着处于本发明的范围之内并且形成不同的实施例。例如,在下面的权利要求书中,所要求保护的实施例的任意之一都可以以任意的组合方式来使用。
应该注意的是上述实施例对本发明进行说明而不是对本发明进行限制,并且本领域技术人员在不脱离所附权利要求的范围的情况下可设计出替换实施例。在权利要求中,不应将位于括号之间的任何参考符号构造成对权利要求的限制。单词“包含”不排除存在未列在权利要求中的元件或步骤。位于元件之前的单词“一”或“一个”不排除存在多个这样的元件。本发明可以借助于包括有若干不同元件的硬件以及借助于适当编程的计算机来实现。在列举了若干装置的单元权利要求中,这些装置中的若干个可以是通过同一个硬件项来具体体现。单词第一、第二、以及第三等的使用不表示任何顺序。可将这些单词解释为名称。上述实施例中的步骤,除有特殊说明外,不应理解为对执行顺序的限定。
Claims (19)
- 一种激光雷达参数调整方法,其特征在于,包括:获取激光雷达周围三维环境信息;基于所述三维环境信息,识别所述激光雷达所处的场景类型以及可行驶区域;根据所述场景类型和所述可行驶区域,确定所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
- 根据权利要求1所述的方法,其特征在于,所述三维环境信息包括三维点云数据;所述基于所述三维环境信息,识别所述激光雷达所处的场景类型以及可行驶区域,具体包括:对所述三维点云数据进行处理,生成多通道点云特征图;从所述多通道点云特征图提取高维特征信息;根据所述高维特征信息,确定所述场景类型;根据所述高维特征信息,确定所述可行驶区域。
- 根据权利要求2所述的方法,其特征在于,所述从所述多通道点云特征图提取高维特征信息,具体包括:将所述多通道点云特征图输入第一神经网络,并获取所述第一神经网络输出的所述高维特征信息;所述根据所述高维特征信息,确定所述场景类型,具体包括:将所述高维特征信息输入第二神经网络,并获取所述第二神经网络输出的场景类型输出值;根据所述场景类型输出值与场景类型标签的对应关系,确定所述场景类型输出值对应的所述场景类型;所述根据所述高维特征信息,确定所述可行驶区域,具体包括:将所述高维特征信息输入第三神经网络,并获取所述第三神经网络输出的可行驶区域输出图;根据预设可行驶区域输出图与可行驶区域标签的对应关系,确定所述可行驶区域输出图对应的所述可行驶区域。
- 根据权利要求1所述的方法,其特征在于,所述根据所述场景类型和所述可行驶区域,确定所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数,具体包括:根据所述场景类型,确定所述激光雷达的水平视场角、所述激光雷达的 垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略;根据所述可行驶区域,调整所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
- 根据权利要求4所述的方法,其特征在于,所述根据所述可行驶区域,调整所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数,具体包括:根据所述场景类型,获取所述场景类型下的标准行驶区域以及所述标准行驶区域对应的所述激光雷达的水平视场角以及所述激光雷达的垂直视场角的行驶参数调整策略;将所述可行驶区域与标准行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,对所述场景参数调整策略进行再调整;基于调整后的所述场景参数调整策略调整所述激光雷达的当前工作参数。
- 根据权利要求5所述的方法,其特征在于,所述将所述标准行驶区域和所述可行驶区域进行匹配,根据所述匹配的结果以及所述行驶参数调整策略,调整所述场景参数调整策略,具体包括:分析所述可行驶区域的拓扑结构以及所述标准行驶区域的拓扑结构;计算所述可行驶区域的拓扑结构和所述标准行驶区域的拓扑结构的拓扑相关性;根据所述拓扑相关性,确定所述可行驶区域相对于所述标准行驶区域的场景差异系数;根据所述场景差异系数以及所述行驶参数调整策略,调整所述场景参数调整策略。
- 根据权利要求1-6任一项所述的方法,其特征在于,所述激光雷达的当前工作参数包括:所述激光雷达的发射光束的垂直视场角、所述激光雷达的发射光束的水平视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描频率以及激光雷达的脉冲发射频率中的一种或多种。
- 一种激光雷达参数调整装置,其特征在于,包括:环境信息获取模块,用于获取激光雷达周围三维环境信息;识别模块,用于基于所述三维环境信息,识别所述激光雷达所处的场景类型以及可行驶区域;调整模块,用于根据所述场景类型和所述可行驶区域,确定所述激光雷达的参数调整策略,并基于所述参数调整策略调整所述激光雷达的当前工作参数。
- 一种激光雷达,其特征在于,包括:发射装置、接收装置、处理器、存储器、通信接口和通信总线,所述处理器、所述存储器和所述通信接口通过所述通信总线完成相互间的通信;所述发射装置用于发射出射激光至探测区域;所述接收装置用于接收所述探测区域内的物体反射的回波激光;所述存储器用于存放至少一可执行指令,所述可执行指令使所述处理器执行根据权利要求1-7任一项所述的激光雷达参数调整方法的步骤,对所述发射装置的工作参数进行调整。
- 一种激光雷达参数调整方法,应用于包括多个激光雷达的激光雷达***,其特征在于,包括:获取所述激光雷达***周围三维环境信息;基于所述三维环境信息,识别所述激光雷达***所处的场景类型以及可行驶区域;根据所述场景类型,确定所述激光雷达***中至少一个所述激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略;根据所述可行驶区域,调整至少一个所述激光雷达的所述场景参数调整策略,并将调整后的场景参数调整策略确定为至少一个所述激光雷达的参数调整策略,并基于所述参数调整策略调整至少一个所述激光雷达的当前工作参数。
- 根据权利要求10所述的方法,其特征在于,所述获取所述激光雷达***周围三维环境信息,包括:通过所述激光雷达***中的一个激光雷达获取所述激光雷达***周围三维环境信息;其中,所述一个激光雷达为探测中远探测视场的激光雷达;或者,通过所述激光雷达***中的多个激光雷达分别获取所述激光雷达***周围三维环境信息,其中,所述多个激光雷达对应不同的探测视场,整合所述多个激光雷达分别获取的所述三维环境信息,得到所述激光雷达***完整的三维环境信息。
- 根据权利要求10所述的方法,其特征在于,基于所述三维环境信息,识别所述激光雷达***所处的场景类型以及可行驶区域之后,所述方法还包括:根据所述场景类型确定与所述场景类型对应的扫描区域范围;确定激光雷达***中各个激光雷达的扫描区域范围;根据所述场景类型对应的扫描区域范围和所述激光雷达***中各个激光雷达的扫描区域范围确定所述激光雷达***中处于工作状态的激光雷达。
- 根据权利要求12所述的方法,其特征在于,所述根据所述场景类型对应的扫描区域范围和所述激光雷达***中各个激光雷达的扫描区域范围确定所述激光雷达***中处于工作状态的激光雷达,包括:当识别的所述场景类型为高速公路的场景时,选择所述激光雷达***中用于探测中远探测视场的激光雷达工作;当识别的所述场景类型为十字路口场景时,选择激光雷达***中所有激光雷达工作。
- 根据权利要求10所述的方法,其特征在于,所述根据所述场景类型,确定所述激光雷达***中至少一个所述激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略,包括:获取所述激光雷达***中的各个激光雷达的标识;根据预设的所述场景类型、所述激光雷达***中至少一个所述激光雷达的标识和所述场景参数调整策略的对应关系,确定与所述场景类型以及所述激光雷达***中至少一个激光雷达的标识对应的所述激光雷达***中至少一个激光雷达的水平视场角、所述激光雷达的垂直视场角、所述激光雷达的发射光束的光轴方向、所述激光雷达的扫描密度、所述激光雷达的扫描频率以及所述激光雷达的脉冲发射功率中的一种或多种的场景参数调整策略。
- 根据权利要求10所述的方法,其特征在于,所述根据所述可行驶区域,调整至少一个所述激光雷达的所述场景参数调整策略,并将所述调整后的场景参数调整策略确定为至少一个所述激光雷达的参数调整策略,并基于所述参数调整策略调整至少一个所述激光雷达的当前工作参数,具体包括:根据所述场景类型,获取所述场景类型下的标准行驶区域以及所述标准行驶区域对应的所述激光雷达***中至少一个所述激光雷达的水平视场角以及垂直视场角的行驶参数调整策略;将所述标准行驶区域和所述可行驶区域进行匹配,根据匹配的结果以及所述激光雷达***中至少一个所述激光雷达的行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的场景参数调整策略;基于所述激光雷达***中至少一个所述激光雷达的场景参数调整策略调整所述激光雷达***中至少一个所述激光雷达的当前工作参数。
- 根据权利要求15所述的方法,其特征在于,所述将所述标准行驶区域和所述可行驶区域进行匹配,根据匹配的结果以及所述激光雷达***中至少一个所述激光雷达的行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的场景参数调整策略,包括:分析所述可行驶区域的拓扑结构以及所述标准行驶区域的拓扑结构;计算所述可行驶区域的拓扑结构和所述标准行驶区域的拓扑结构的拓扑相关性;根据所述拓扑相关性,确定所述可行驶区域相对于所述标准行驶区域的场景差异系数;根据所述场景差异系数以及所述激光雷达***中至少一个所述激光雷达的所述行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的所述场景参数调整策略。
- 根据权利要求16所述的方法,其特征在于,所述根据所述场景差异系数以及所述激光雷达***中至少一个所述激光雷达的所述行驶参数调整策略,调整所述激光雷达***中至少一个所述激光雷达的所述场景参数调整策略,包括:获取所述激光雷达***中至少一个所述激光雷达的所述行驶参数调整策略中的标准视场角参数;根据所述场景差异系数,在所述标准视场角参数的基础上等比例计算新的视场角参数;根据所述新的视场参数,调整所述激光雷达***中至少一个所述激光雷达的所述场景参数调整策略中的垂直视场角和水平视场角。
- 一种激光雷达***,其特征在于,包括多个激光雷达,所述多个激光雷达设于激光雷达***承载装置的同一平面上,或者设于所述激光雷达***承载装置的不同平面上,所述激光雷达包括:发射装置、接收装置、处理器、存储器、通信接口和通信总线,所述处理器、所述存储器和所述通信接口通过所述通信总线完成相互间的通信;所述发射装置用于发射出射激光至探测区域;所述接收装置用于接收所述探测区域内的物体反射的回波激光;所述存储器用于存放至少一可执行指令,所述可执行指令使所述处理器执行根据权利要求10-17任一项所述的激光雷达参数调整方法的步骤,对所述发射装置的工作参数进行调整。
- 一种计算机存储介质,其特征在于,所述存储介质中存储有至少一可执行指令,所述可执行指令使处理器执行根据权利要求1-7和10-17任一项所述的激光雷达参数调整方法的步骤。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US18/094,346 US11768293B2 (en) | 2020-07-08 | 2023-01-07 | Method and device for adjusting parameters of LiDAR, and LiDAR |
US18/227,283 US20240027621A1 (en) | 2020-07-08 | 2023-07-27 | Method and device for adjusting parameters of lidar, and lidar |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010650848.4 | 2020-07-08 | ||
CN202010650848.4A CN111537980B (zh) | 2020-07-08 | 2020-07-08 | 激光雷达参数调整方法、装置及激光雷达 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US18/094,346 Continuation US11768293B2 (en) | 2020-07-08 | 2023-01-07 | Method and device for adjusting parameters of LiDAR, and LiDAR |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022007859A1 true WO2022007859A1 (zh) | 2022-01-13 |
Family
ID=71980901
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2021/105048 WO2022007859A1 (zh) | 2020-07-08 | 2021-07-07 | 激光雷达参数调整方法、装置及激光雷达 |
Country Status (3)
Country | Link |
---|---|
US (2) | US11768293B2 (zh) |
CN (2) | CN111537980B (zh) |
WO (1) | WO2022007859A1 (zh) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111537980B (zh) * | 2020-07-08 | 2020-11-27 | 深圳市速腾聚创科技有限公司 | 激光雷达参数调整方法、装置及激光雷达 |
CN112748417B (zh) * | 2020-12-28 | 2023-06-06 | 北京一径科技有限公司 | 激光雷达控制方法及装置、电子设备及存储介质 |
CN113178091B (zh) * | 2021-05-12 | 2022-06-10 | 中移智行网络科技有限公司 | 安全行驶区域方法、装置和网络设备 |
CN113806464A (zh) * | 2021-09-18 | 2021-12-17 | 北京京东乾石科技有限公司 | 路牙确定方法、装置、设备以及存储介质 |
CN114325730B (zh) * | 2021-11-17 | 2022-12-13 | 杭州宏景智驾科技有限公司 | 一种大视场远距离激光雷达测试方法 |
CN114488191A (zh) * | 2022-01-13 | 2022-05-13 | 杭州涂鸦信息技术有限公司 | 激光雷达扫描方法、控制设备及激光雷达 |
CN114571450A (zh) * | 2022-02-23 | 2022-06-03 | 达闼机器人股份有限公司 | 机器人控制方法、装置及存储介质 |
CN114966620B (zh) * | 2022-04-18 | 2023-05-05 | 深圳阜时科技有限公司 | 光电检测装置及电子设备 |
CN114637016B (zh) * | 2022-05-11 | 2022-08-09 | 西安晟昕科技发展有限公司 | 一种激光雷达信号发射及信号处理方法 |
CN115291198B (zh) * | 2022-10-10 | 2023-01-24 | 西安晟昕科技发展有限公司 | 一种雷达信号发射及信号处理方法 |
US11828846B1 (en) * | 2023-06-10 | 2023-11-28 | IntraAction Corp. | Lidar acousto-optic device and methods of fabrication |
CN116559896B (zh) * | 2023-07-10 | 2023-10-27 | 深圳市欢创科技有限公司 | 调整激光雷达测距精度的方法、装置及激光雷达 |
CN117554992B (zh) * | 2024-01-10 | 2024-04-09 | 青岛镭测创芯科技有限公司 | 基于激光雷达的消光系数获取方法及*** |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160146941A1 (en) * | 2014-11-24 | 2016-05-26 | Conti Temic Microelectronic Gmbh | Vehicle Environment Scanning by a Phase-Controlled Laser |
CN108375775A (zh) * | 2018-01-17 | 2018-08-07 | 上海禾赛光电科技有限公司 | 车载探测设备及其参数的调整方法、介质、探测*** |
CN109543600A (zh) * | 2018-11-21 | 2019-03-29 | 成都信息工程大学 | 一种实现可行驶区域检测方法及***和应用 |
CN109597093A (zh) * | 2018-11-14 | 2019-04-09 | 深圳市速腾聚创科技有限公司 | 激光雷达的参数调整方法及装置 |
CN110275167A (zh) * | 2019-06-03 | 2019-09-24 | 浙江吉利控股集团有限公司 | 一种雷达探测的控制方法、控制器及终端 |
CN111537980A (zh) * | 2020-07-08 | 2020-08-14 | 深圳市速腾聚创科技有限公司 | 激光雷达参数调整方法、装置及激光雷达 |
Family Cites Families (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE10360890A1 (de) * | 2003-12-19 | 2005-07-21 | Robert Bosch Gmbh | Radarsensor und Verfahren zu dessen Betrieb |
CN103760569B (zh) * | 2013-12-31 | 2016-03-30 | 西安交通大学 | 一种基于激光雷达的可行驶区域检测方法 |
US9715016B2 (en) * | 2015-03-11 | 2017-07-25 | The Boeing Company | Real time multi dimensional image fusing |
CN108072880A (zh) * | 2018-01-17 | 2018-05-25 | 上海禾赛光电科技有限公司 | 激光雷达视场中心指向的调整方法、介质、激光雷达*** |
US11221413B2 (en) * | 2018-03-14 | 2022-01-11 | Uatc, Llc | Three-dimensional object detection |
CN110531339B (zh) * | 2018-05-25 | 2021-08-03 | 北京万集科技股份有限公司 | 激光雷达的探测方法、装置及计算机可读存储介质 |
CN108764187B (zh) * | 2018-06-01 | 2022-03-08 | 百度在线网络技术(北京)有限公司 | 提取车道线的方法、装置、设备、存储介质以及采集实体 |
CN110654320A (zh) * | 2018-06-29 | 2020-01-07 | 比亚迪股份有限公司 | 探测装置探测角度的调整方法、装置及具有其的车辆 |
CN110376594B (zh) * | 2018-08-17 | 2022-02-01 | 北京京东叁佰陆拾度电子商务有限公司 | 一种基于拓扑图的智能导航的方法和*** |
CN109143207B (zh) * | 2018-09-06 | 2020-11-10 | 百度在线网络技术(北京)有限公司 | 激光雷达内参精度验证方法、装置、设备及介质 |
CN109871850A (zh) * | 2019-01-21 | 2019-06-11 | 北京大学 | 一种基于神经网络模型的车载激光雷达数据的分类方法 |
CN109884651B (zh) * | 2019-04-03 | 2023-04-18 | 深圳市镭神智能***有限公司 | 一种激光雷达扫描方法和激光雷达 |
CN110171417B (zh) * | 2019-05-24 | 2020-08-11 | 无锡佶达德光电子技术有限公司 | 基于激光雷达的定速巡航装置 |
CN110824476A (zh) * | 2019-08-26 | 2020-02-21 | 惠州市德赛西威汽车电子股份有限公司 | 一种探测范围动态可调的汽车探测方法及雷达*** |
CN110795821B (zh) * | 2019-09-25 | 2024-04-09 | 的卢技术有限公司 | 一种基于场景区分的深度强化学习训练方法及*** |
CN110703216B (zh) * | 2019-10-21 | 2023-08-22 | 苏州豪米波技术有限公司 | 一种雷达调整***、调整方法及雷达装置 |
CN110703223B (zh) * | 2019-11-07 | 2023-06-30 | 上海禾赛科技有限公司 | 应用于激光雷达的调节方法和电子设备 |
-
2020
- 2020-07-08 CN CN202010650848.4A patent/CN111537980B/zh active Active
- 2020-07-08 CN CN202011090310.9A patent/CN111999720B/zh active Active
-
2021
- 2021-07-07 WO PCT/CN2021/105048 patent/WO2022007859A1/zh active Application Filing
-
2023
- 2023-01-07 US US18/094,346 patent/US11768293B2/en active Active
- 2023-07-27 US US18/227,283 patent/US20240027621A1/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160146941A1 (en) * | 2014-11-24 | 2016-05-26 | Conti Temic Microelectronic Gmbh | Vehicle Environment Scanning by a Phase-Controlled Laser |
CN108375775A (zh) * | 2018-01-17 | 2018-08-07 | 上海禾赛光电科技有限公司 | 车载探测设备及其参数的调整方法、介质、探测*** |
CN109597093A (zh) * | 2018-11-14 | 2019-04-09 | 深圳市速腾聚创科技有限公司 | 激光雷达的参数调整方法及装置 |
CN109543600A (zh) * | 2018-11-21 | 2019-03-29 | 成都信息工程大学 | 一种实现可行驶区域检测方法及***和应用 |
CN110275167A (zh) * | 2019-06-03 | 2019-09-24 | 浙江吉利控股集团有限公司 | 一种雷达探测的控制方法、控制器及终端 |
CN111537980A (zh) * | 2020-07-08 | 2020-08-14 | 深圳市速腾聚创科技有限公司 | 激光雷达参数调整方法、装置及激光雷达 |
CN111999720A (zh) * | 2020-07-08 | 2020-11-27 | 深圳市速腾聚创科技有限公司 | 激光雷达参数调整方法、激光雷达***和计算机存储介质 |
Also Published As
Publication number | Publication date |
---|---|
CN111537980A (zh) | 2020-08-14 |
CN111999720A (zh) | 2020-11-27 |
US20230161045A1 (en) | 2023-05-25 |
US11768293B2 (en) | 2023-09-26 |
CN111537980B (zh) | 2020-11-27 |
US20240027621A1 (en) | 2024-01-25 |
CN111999720B (zh) | 2022-08-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2022007859A1 (zh) | 激光雷达参数调整方法、装置及激光雷达 | |
US20220043449A1 (en) | Multi-channel sensor simulation for autonomous control systems | |
US11821988B2 (en) | Ladar system with intelligent selection of shot patterns based on field of view data | |
US11885910B2 (en) | Hybrid-view LIDAR-based object detection | |
US11593950B2 (en) | System and method for movement detection | |
Caltagirone et al. | Fast LIDAR-based road detection using fully convolutional neural networks | |
Chen et al. | Gaussian-process-based real-time ground segmentation for autonomous land vehicles | |
Prophet et al. | Semantic segmentation on automotive radar maps | |
CN115151467A (zh) | 用于成像***的车道检测和跟踪技术 | |
US11798289B2 (en) | Streaming object detection and segmentation with polar pillars | |
KR20230026130A (ko) | 자율 주행을 위한 단일 계층 3차원 다중 객체 검출 장치 및 방법 | |
Stroescu et al. | Object detection on radar imagery for autonomous driving using deep neural networks | |
Bruno et al. | Analysis and fusion of 2d and 3d images applied for detection and recognition of traffic signs using a new method of features extraction in conjunction with deep learning | |
Wang et al. | A new scheme of vehicle detection for severe weather based on multi-sensor fusion | |
CN117237919A (zh) | 跨模态监督学习下多传感器融合检测的卡车智驾感知方法 | |
CN113888463B (zh) | 车轮转角的检测方法、装置、电子设备及存储介质 | |
Rana et al. | Comparative study of Automotive Sensor technologies used for Unmanned Driving | |
EP4170606A1 (en) | Identification of real and image sign detections in driving applications | |
US20220371606A1 (en) | Streaming object detection and segmentation with polar pillars | |
Ponnaganti et al. | Utilizing CNNs for Object Detection with LiDAR Data for Autonomous Driving | |
Zheng | (Retracted) Laser radar-based intelligent vehicle target recognition and detection system using image detection technology | |
Nayak et al. | BEV detection and localisation using semantic segmentation in autonomous car driving systems | |
Rui et al. | A study for obstacle determination fusing mmw radar and vision sensor | |
TWI832686B (zh) | 路徑規劃系統及其路徑規劃方法 | |
Carvalho et al. | Exploring Millimeter-Wave Radar Algorithms for Obstacle Detection and Tracking |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21838461 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 18/04/2023) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21838461 Country of ref document: EP Kind code of ref document: A1 |