WO2021042286A1 - 点云处理方法、***、可移动平台及存储介质 - Google Patents

点云处理方法、***、可移动平台及存储介质 Download PDF

Info

Publication number
WO2021042286A1
WO2021042286A1 PCT/CN2019/104366 CN2019104366W WO2021042286A1 WO 2021042286 A1 WO2021042286 A1 WO 2021042286A1 CN 2019104366 W CN2019104366 W CN 2019104366W WO 2021042286 A1 WO2021042286 A1 WO 2021042286A1
Authority
WO
WIPO (PCT)
Prior art keywords
point clouds
adjacent
positional relationship
relative positional
aggregated
Prior art date
Application number
PCT/CN2019/104366
Other languages
English (en)
French (fr)
Inventor
朱振宇
Original Assignee
深圳市大疆创新科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市大疆创新科技有限公司 filed Critical 深圳市大疆创新科技有限公司
Priority to CN201980034379.9A priority Critical patent/CN112166460A/zh
Priority to PCT/CN2019/104366 priority patent/WO2021042286A1/zh
Publication of WO2021042286A1 publication Critical patent/WO2021042286A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G06T17/05—Geographic models
    • 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
    • G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
    • G01S17/50—Systems of measurement based on relative movement of target
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
    • G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle

Definitions

  • the embodiments of the present application relate to the field of detection technology, and in particular, to a point cloud processing method, system, removable platform, and storage medium.
  • a large number of point clouds need to be collected by a detection device.
  • the detection device can collect different point clouds at different times.
  • the relative position relationship between the point cloud collected at each time and the point cloud collected at other times is further calculated, and an electronic map is established according to the relative position relationship between the point cloud at each time and the point cloud at other times.
  • the relative position relationship between adjacent point clouds in time series can be used to solve the relative position relationship between non-adjacent point clouds in time series.
  • the time span between the two sets of point clouds that are not adjacent in time series is relatively long, errors will accumulate, and the relative positional relationship between the point clouds at different times will be inaccurate.
  • the embodiments of the present application provide a point cloud processing method, system, movable platform, and storage medium, so as to improve the accuracy of the relative positional relationship between the point clouds detected by the detection device at different times.
  • the first aspect of the embodiments of the present application is to provide a point cloud processing method, which is applied to a movable platform, the movable platform is provided with a detection device, and the detection device is used to detect and obtain a point cloud, and the method includes:
  • the point clouds detected by the detection device are aggregated to obtain a plurality of aggregated point clouds, and each aggregated point cloud includes at least two points. Describe adjacent point clouds in time series;
  • the relative positional relationship of the target between the point clouds detected by the detection device at different times is determined.
  • the second aspect of the embodiments of the present application is to provide a point cloud processing system, including: a detection device, a memory, and a processor;
  • the detection device is used to detect and obtain a point cloud
  • the memory is used to store program codes
  • the processor calls the program code, and when the program code is executed, is used to perform the following operations:
  • the point clouds detected by the detection device are aggregated to obtain a plurality of aggregated point clouds, and each aggregated point cloud includes at least two points. Describe adjacent point clouds in time series;
  • the relative positional relationship of the target between the point clouds detected by the detection device at different times is determined.
  • the third aspect of the embodiments of the present application is to provide a movable platform, including:
  • the power system is installed on the fuselage to provide mobile power
  • the fourth aspect of the embodiments of the present application is to provide a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the method described in the first aspect.
  • the point cloud processing method, system, movable platform, and storage medium provided in this embodiment can aggregate the adjacent point clouds in time series by obtaining the relative positional relationship between adjacent point clouds in time series to obtain multiple points.
  • a large point cloud is multiple aggregated point clouds. Because the detailed features in the large point cloud are more abundant, even if the adjacent area is blocked by the obstruction, the detailed features in the large point cloud can be used to complete the large point cloud According to the relative position relationship between the large point clouds that are adjacent to each other, the relative position relationship of the target between the point clouds detected by the detection device at different times can be determined more accurately, thereby improving the relative position relationship between the point clouds at different times. The accuracy of the relative positional relationship between the point clouds.
  • FIG. 1 is a schematic diagram of an application scenario provided by an embodiment of the application
  • FIG. 2 is a flowchart of a point cloud processing method provided by an embodiment of the application
  • FIG. 3 is a schematic diagram of a point cloud aggregation provided by an embodiment of the application.
  • FIG. 4 is a flowchart of a point cloud processing method provided by another embodiment of this application.
  • Fig. 5 is a structural diagram of a point cloud processing system provided by an embodiment of the application.
  • Fig. 6 is a schematic diagram of a point cloud provided by an embodiment of the application.
  • a component when referred to as being "fixed to” another component, it can be directly on the other component or a central component may also exist. When a component is considered to be “connected” to another component, it can be directly connected to the other component or there may be a centered component at the same time.
  • the embodiment of the present application provides a point cloud processing method.
  • the method can be applied to a movable platform, and the movable platform is provided with a detection device, and the detection device is used to detect and obtain a point cloud.
  • the movable platform may be a drone, a movable robot or a vehicle.
  • the movable platform is a vehicle as an example.
  • the vehicle may be an unmanned vehicle or a vehicle equipped with an Advanced Driver Assistance Systems (ADAS) system.
  • ADAS Advanced Driver Assistance Systems
  • the vehicle 11 is a carrier equipped with a detection device.
  • the detection device may specifically be a binocular stereo camera, a time of flight (TOF) camera, a millimeter wave radar, and/or a lidar.
  • TOF time of flight
  • the detection device detects objects around the vehicle 11 in real time to obtain a point cloud.
  • the point cloud may also be referred to as a three-dimensional point cloud. Take lidar as an example.
  • the lidar can determine the relative position of the object based on the laser light reflected from the surface of the object. Information such as the position and distance of the lidar. If the laser beam emitted by the lidar is scanned according to a certain trajectory, such as a 360-degree rotating scan, a large number of laser points will be obtained, and thus the laser point cloud data of the object can be formed, that is, a point cloud.
  • the point cloud processing method can be executed by the vehicle-mounted device in the vehicle, or can be executed by other devices with data processing functions besides the vehicle-mounted device, for example, such as The server 12 shown in FIG. 1, the vehicle 11 and the server 12 may perform wireless communication or wired communication.
  • the vehicle 11 may send the point cloud detected by the detection device to the server 12, and the server 12 executes the point cloud processing method.
  • an in-vehicle device is used as an example to introduce the point cloud processing method provided in the embodiment of the present application.
  • the vehicle-mounted device may be a device with a data processing function integrated in the vehicle center console, or may also be a tablet computer, a mobile phone, a notebook computer, etc. placed in the vehicle.
  • Fig. 2 is a flowchart of a point cloud processing method provided by an embodiment of the application. As shown in Figure 2, the method in this embodiment may include:
  • Step S201 Obtain the relative positional relationship between adjacent point clouds in time series.
  • the detection device mounted on the vehicle 11 detects objects around the vehicle 11 in real time to obtain a point cloud.
  • the detection device can communicate with the on-board equipment on the vehicle 11, so that the vehicle 11
  • the on-board equipment can obtain the point cloud detected by the detection equipment in real time. For example, at time t1, the point cloud acquired by the vehicle-mounted device is recorded as P 1 ; at time t2, the point cloud acquired by the vehicle-mounted device is recorded as P 2 ; and so on, at time ti, the point cloud acquired by the vehicle-mounted device is recorded as P 1 Is P i .
  • the point cloud acquired by the vehicle-mounted device may include multiple three-dimensional points, and each three-dimensional point corresponds to a three-dimensional coordinate.
  • t1, t2, ..., ti may be consecutive time points. Therefore, P 1 and P 2 , P 2 and P 3 , ..., P i-1 and P i can be regarded as point clouds adjacent in time series. It is understandable that, compared to non-adjacent point clouds in time series, the overlap area between adjacent point clouds in time series is relatively large. Therefore, it can be determined according to the overlap area between adjacent point clouds in time series.
  • the relative positional relationship between point clouds that are not adjacent in time series, and the relative positional relationship may specifically include a rotation relationship and/or a translation relationship.
  • the acquiring the relative positional relationship between adjacent point clouds in time series includes: calculating the relative positional relationship between adjacent point clouds in time series according to an iterated closest point (Iterated ClosestPoints, ICP) algorithm .
  • ICP iterated closest point
  • P 1 and P 2 as an example, p (i) of P. 1 in any point, is determined from p (i) the nearest point m (i), p (i ) and m (i in P 2 ) Can form a set of corresponding point pairs.
  • multiple sets of corresponding point pairs in P 1 and P 2 can be determined, for example, p(1) and m(1), p(2) and m(2) ,..., p(n) and m(n) are n sets of corresponding point pairs.
  • p(1), p(2),..., p(n) are points in the point cloud P 1 respectively
  • m(1), m(2),..., m(n) are the point clouds respectively points 2 P
  • m (1) P 2 is a point cloud distance p (1) nearest one
  • m (2) is a point cloud from p (2) nearest a point P 2, and so on.
  • P 1 represents a point cloud
  • p (1) represents a point P 1 in the point cloud
  • N sets of corresponding point pairs can establish n equations. Further, using mathematical methods to solve the n equations can obtain the rotation and translation relationships between P 1 and P 2.
  • an iterative algorithm can also be used, for example, the ICP algorithm to calculate the rotation relationship and the translation relationship between P 1 and P 2.
  • the specific iterative algorithm can be implemented as follows:
  • a possible implementation is to set the n sets of corresponding point pairs as described above, for example, p(1) and m(1), p(2) and m(2), ..., p(n) and m(n) is substituted into the function E described in formula (1):
  • the function E described in formula (1) is optimized to obtain the rotation relationship R and the translation relationship T that make the value of the function E the smallest or less than a certain threshold.
  • Step S202 Perform aggregation processing on the point clouds detected by the detection device according to the relative positional relationship between adjacent point clouds in the time series to obtain multiple aggregate point clouds, each of which includes at least two aggregate point clouds.
  • the point cloud adjacent in the time series is a point cloud adjacent in the time series.
  • the vehicle 11 drives along the road section 13 and then adjusts the direction to continue to drive along the road section 14, and the road section 13 and the road section 14 are the incoming circuit sections.
  • the road section 13 and the road section 14 can be divided by barriers, green belts, and the like.
  • the point clouds detected by the detection device are P 1 , P 2 , ..., P 20 in order , as shown in FIG. 3, for example.
  • the relative positional relationship between every two adjacent point clouds in time sequence in P 1 , P 2 , ..., P 20 can be calculated, that is, P 1 and P 2 , The relative positional relationship between P 2 and P 3 ,..., P 19 and P 20 , where the rotational relationship between P 1 and P 2 is denoted as R P1, P2 , and the relationship between P 1 and P 2
  • the translation relationship is denoted as T P1, P2
  • the rotation relationship between P 2 and P 3 is denoted as R P2, P3
  • the translation relationship between P 2 and P 3 is denoted as T P2, P3
  • the rotation relationship between P 19 and P 20 is denoted as R P19, P20
  • the translation relationship between P 19 and P 20 is denoted as T P19, P20 .
  • P 1 , P 2 , ..., P 20 can also be aggregated, for example, P 1 , P 2 , P 3 , P 4 , P 5 are aggregated into an aggregate point cloud and denoted as M 1 , P 6 , P 7 , P 8 , P 9 , and P 10 are aggregated into an aggregate point cloud and denoted as M 2 , and P 11 , P 12 ,
  • the aggregation of P 13 , P 14 , and P 15 into an aggregate point cloud is denoted as M 4
  • the aggregation of P 16 , P 17 , P 18 , P 19 , and P 20 into an aggregate point cloud is denoted as M 3 .
  • the number of point clouds included in each aggregate point cloud may be the same or different, and this embodiment does not limit the number of point clouds that each aggregate point cloud may include.
  • each The aggregate point cloud includes at least two time-adjacent point clouds
  • the following possible situations may be included when the multiple point clouds detected by the detection device are aggregated:
  • the performing aggregation processing on the point clouds detected by the detection device according to the relative positional relationship between adjacent point clouds in the time series includes: For the relative positional relationship between adjacent point clouds, the point clouds within the preset distance range are aggregated.
  • P 1 from the beginning, in the order of time, to find the next time point cloud P 2, and P 2 is calculated with respect to the distance P 1 if P 2 P 1 with respect to the distance is less than a predetermined distance threshold, For example, 100 meters, continue to look for the next point cloud P 3 , if the distance of P 3 relative to P 1 is less than the preset distance threshold, continue to look for the next point cloud until a point cloud such as P 6 , P 6 is found If the distance at P 1 is greater than the preset distance threshold, then P 1 , P 2 , P 3 , P 4 , and P 5 are aggregated into an aggregated point cloud M 1 .
  • a predetermined distance threshold For example, 100 meters, continue to look for the next point cloud P 3 , if the distance of P 3 relative to P 1 is less than the preset distance threshold, continue to look for the next point cloud until a point cloud such as P 6 , P 6 is found.
  • the distance between the two point clouds may specifically be the distance between the two point clouds calculated by projecting the two point clouds to the same coordinate system.
  • P 2 from P 1 with respect to an example may be projected point cloud projected onto P 1 P 2 located in the coordinate system, for example, the P 1 P 2 projected coordinate system is obtained where P '2,
  • the distance of P′ 2 relative to P 1 can be regarded as the distance of P 2 relative to P 1 .
  • P '2 with respect to the distance P 1 may be a variety of calculations, e.g., P' 2 P 1 with respect to the distance may be the distance between the center point P 'with the center point P 2 1, will be understood This is just an example, and there may be other different calculation methods in other embodiments.
  • the aggregation processing of point clouds within a preset distance range according to the relative positional relationship between adjacent point clouds in the time series includes: according to the relationship between the adjacent point clouds in the time series The relative position relationship of, converts the point cloud within the preset distance range to the same coordinate system.
  • P 1 , P 2 , P 3 , P 4 , and P 5 can be transformed into the same coordinate system, For example, convert P 2 , P 3 , P 4 , and P 5 to the coordinate system where P 1 is located.
  • P 6 , P 7 , P 8 , P 9 , and P 10 into M 2 P 7 , P 8 , P 9 , and P 10 can be converted to the coordinate system where P 6 is located.
  • P 12 , P 13 , P 14 , and P 15 can be converted to the coordinate system where P 11 is located.
  • P 16 , P 17 , P 18 , P 19 , and P 20 into M 3 P 17 , P 18 , P 19 , and P 20 can be converted to the coordinate system where P 16 is located.
  • the performing aggregation processing on the point cloud detected by the detection device according to the relative position relationship between adjacent point clouds in the time series includes: according to the time series For the relative positional relationship between adjacent point clouds, the point clouds within the preset time range are aggregated.
  • P 1 , P 1 , P 2 , P 3 , P 4 , and P 5 are within the preset time range. For example, if the point cloud is detected within 30 seconds, then P 1 , P 2 , P 3 , P 4 , and P 5 are aggregated into an aggregated point cloud M 1 .
  • P 6 , P 6 , P 7 , P 8 , P 9 , P 10 are the point clouds detected in the preset time range, then P 6 , P 7 , P 8 , P 9 , P 10 Aggregate as M 2 .
  • P 11 , P 11 , P 12 , P 13 , P 14 , P 15 are the point clouds detected in the preset time range, then P 11 , P 12 , P 13 , P 14 , P 15 are aggregated into M 4 .
  • P 16 , P 16 , P 17 , P 18 , P 19 , P 20 are the point clouds detected in the preset time range, then P 16 , P 17 , P 18 , P 19 , P 20 are aggregated into M 3 .
  • the aggregation processing of the point clouds within a preset time range according to the relative positional relationship between the adjacent point clouds in the time series includes: according to the relationship between the adjacent point clouds in the time series The relative position relationship of, converts the point cloud within the preset time range into the same coordinate system.
  • P 1 , P 2 , P 3 , P 4 , and P 5 can be transformed into the same coordinate system, For example, convert P 2 , P 3 , P 4 , and P 5 to the coordinate system where P 1 is located.
  • the conversion matrix is the relative position relationship as described above.
  • K 1,3 represents the conversion matrix between P 3 and P 1
  • K 1,3 can be obtained from K 1, 2 and K 2,3
  • K 2,3 represents the conversion matrix between P 3 and P 2
  • K 2,3 P 2 and P may be a relationship between the rotation 3 R P2, P3, and P 2 and the translation relationship between the 3 P T P2, P3 according to obtain.
  • K 1,4 represents the conversion matrix between P 4 and P 1
  • K 1,4 can be obtained from K 1,2 , K 2,3 , K 3,4
  • K 3,4 represents P 4 and P
  • the conversion matrix between 3 , K 3 , 4 can be obtained according to the rotation relationship and the translation relationship between P 4 and P 3.
  • K 1,5 represents the conversion matrix between P 5 and P 1
  • K 1,5 can be obtained according to K 1,2 , K 2,3 , K 3,4 , K 4,5 , K 4,5 represents P 4
  • the conversion matrix between P and P 5 , K 4,5 can be obtained according to the rotation relationship and translation relationship between P 4 and P 5. That is, M 1 is related to the conversion matrix between point clouds adjacent in time series among P 1 , P 2 , P 3 , P 4 , and P 5.
  • M 2 P 6 +K 6,7 P 7 +K 6,8 P 8 +K 6,9 P 9 +K 6,10 P 10 , where K 6,7 represents P
  • K 6,8 represents the conversion matrix between P 8 and P 6
  • K 6,9 represents the conversion matrix between P 9 and P 6
  • K 6,10 represents P 10 and transformation matrix between 6 P.
  • K 6,7 , K 6,8 , K 6,9 , K 6,10 is similar to the calculation process of K 1,2 , K 1,3 , K 1,4 , K 1,5 as described above , I won’t repeat it here. That is, M 2 is related to the conversion matrix between point clouds adjacent in time series in P 6 , P 7 , P 8 , P 9 , and P 10.
  • M 4 P 11 +K 11,12 P 12 +K 11,13 P 13 +K 11,14 P 14 +K 11,15 P 15 , where K 11,12 , K
  • the meaning and calculation process of 11,13 , K 11,14 , K 11,15 are similar to the meaning and calculation process of K 1,2 , K 1,3 , K 1,4 , K 1,5 as described above. I won't repeat them here. That is, M 4 is related to the conversion matrix between point clouds adjacent in time series among P 11 , P 12 , P 13 , P 14 , and P 15.
  • M 3 P 16 +K 16,17 P 17 +K 16,18 P 18 +K 16,19 P 19 +K 16,20 P 20 .
  • the meaning and calculation process of K 16,17 , K 16,18 , K 16,19 , K 16,20 are the same as those of K 1,2 , K 1,3 , K 1,4 , K 1, The meaning and calculation process of 5 are similar, so I won't repeat them here. That is, M 3 is related to the conversion matrix between point clouds adjacent in time series in P 16 , P 17 , P 18 , P 19 , and P 20.
  • Step S203 Determine the relative position relationship between the adjacent aggregate point clouds among the multiple aggregate point clouds.
  • the adjacent aggregated point clouds in the multiple aggregated point clouds can be determined.
  • the point cloud where the GPS information of the aggregated point cloud may be the GPS information of the intermediate point cloud in the aggregated point cloud, and the GPS information of the intermediate point cloud may specifically be the GPS information of the vehicle when the detection device detects the intermediate point cloud.
  • the cloud point of polymerization P 3 M 1 is an intermediate point cloud
  • the detection device in detecting P 3 may be the GPS time information of the vehicle as a polymerization cloud point GPS information M 1.
  • P 8 is a polymerization intermediate point cloud point cloud M 2
  • the detection device detecting P 8 in the GPS time information of the vehicle can be used as the GPS information aggregation point cloud M 2.
  • P 13 is the intermediate point cloud in the aggregate point cloud M 4
  • the GPS information of the vehicle when the detection device obtains P 13 can be used as the GPS information of the aggregate point cloud M 4 .
  • P 18 is the intermediate point cloud in the aggregated point cloud M 3
  • the GPS information of the vehicle when the detection device obtains P 18 can be used as the GPS information of the aggregated point cloud M 3 .
  • M 1 and M 2 it is determined that the distance between M 1 and M 2 is within a preset distance range, for example, within 10 meters, then M 1 and M 2 are regarded as adjacent to each other.
  • M 2 and M 4 are adjacent aggregation point clouds
  • M 3 and M 4 are adjacent aggregation point clouds
  • M 3 and M 1 are adjacent aggregation points. cloud.
  • calculating the relative positional relationship between M 1 and M 2 i.e., the relationship between the rotation and the translation relationship between M 1 and M 2
  • calculating the relative positional relationship between M 2 and M 4, M 4 and M's. 3 The relative positional relationship between M 3 and M 1 .
  • the above-mentioned ICP algorithm can be used to calculate the relative position relationship between adjacent aggregate point clouds.
  • the specific calculation process can refer to the above-mentioned calculation process of the relative position relationship between P 1 and P 2 , I won’t repeat it here.
  • the rotation relationship between M 1 and M 2 is denoted as R M1, M2
  • the translation relationship between M 1 and M 2 is denoted as T M1, M2
  • the rotation relationship between M 2 and M 4 is denoted as R M2 , M4
  • the translation relationship between M 2 and M 4 is denoted as T M2, M4
  • the rotation relationship between M 3 and M 4 is denoted as R M4, M3
  • the translation relationship between M 3 and M 4 is denoted as T M4, M3
  • the rotation relationship between M 3 and M 1 is denoted as R M3, M1
  • the translation relationship between M 3 and M 1 is denoted as T M3, M1 .
  • the above positioning module is not limited to GPS, and can also be Beidou, Galileo, GLONASS, and so on.
  • Step S204 Determine the target relative position relationship between the point clouds detected by the detection device at different moments according to the relative position relationship between the aggregated point clouds that are adjacent to each other.
  • the conversion matrix D 1,2 may be a conversion matrix from M 2 to M 1 .
  • M 2 and M 4 is rotated relationship between R M2, M4, and the translation relationship between 4 and M 2 M T M2, M4, D 2 can be obtained between the transition matrix M 2 and M 4, 4.
  • the conversion matrix D 2,4 may be a conversion matrix from M 4 to M 2 .
  • M 3 and M according to the relationship between the rotational 4 R M4, M3,.
  • the conversion matrix D 4,3 may be a conversion matrix from M 3 to M 4 .
  • M 3 and M is rotated in accordance with the relationship between the 1 R M3, M1, and M 3 and translational relationships between 1 M T M3, M1, M 3 and M obtained conversion matrix between 1 D 3,1, the
  • the conversion matrix D 3,1 may be a conversion matrix from M 1 to M 3 .
  • the conversion matrix is the relative position relationship as described above.
  • the conversion matrix D 1,2 between M 1 and M 2 is specifically the conversion matrix K 1,6 between P 6 and P 1 .
  • M 1 is related to the conversion matrix between adjacent point clouds in time sequence among P 1 , P 2 , P 3 , P 4 , and P 5
  • M 2 is related to P 6 , P 7 , P 8 , P 9 , P 10
  • the conversion matrix between adjacent point clouds in time series is related, therefore, e 1,2 is related to P 1 , P 2 , P 3 , P 4 , P 5 , P 6 , P 7 , P 8 , P 9 , P In 10 , the conversion matrix between adjacent point clouds in time series is related.
  • M' 2 D 2,4 M 4
  • M 2 is related to the conversion matrix between adjacent point clouds in the sequence of P 6 , P 7 , P 8 , P 9 , and P 10
  • M 4 is related to P 11 , P 12 , P 13 , P 14 , P
  • the conversion matrix between adjacent point clouds in time sequence in 15 is related, therefore, e 2,4 is related to P 6 , P 7 , P 8 , P 9 , P 10 , P 11 , P 12 , P 13 , P 14 ,
  • the conversion matrix between adjacent point clouds in time series in P 15 is related.
  • M' 4 D 4,3 M 3
  • M 4 and M ' may be present between
  • e 4,3 is related to the conversion matrix between adjacent point clouds in time sequence among P 11 , P 12 , P 13 , P 14 , P 15 , P 16 , P 17 , P 18 , P 19 , and P 20.
  • the objective function can be obtained Among them, the objective function F and P 1 , P 2 , P 3 , P 4 , P 5 , P 6 , P 7 , P 8 , P 9 , P 10 , P 11 , P 12 , P 13 , P 14 , P 15 , P 16 , P 17 , P 18 , P 19 , P 20 , the conversion matrix between adjacent point clouds in time series is related.
  • To optimize that is, to solve the minimum function value of the objective function, it can be used for the above-mentioned P 1 and P 2 , P 2 and P 3 ,..., P 19 and P 20 and other adjacent point clouds in time series.
  • the relative position relationship is optimized, that is, the optimized relative position relationship between P 1 and P 2 , P 2 and P 3 ,..., P 19 and P 20 can minimize the function value of the objective function F, Or make the function value of the objective function F less than or equal to the preset threshold.
  • step S202 to step S204 repeating step S202 to step S204 as described above, it can be understood that this The embodiment does not limit the number of repeated executions.
  • each repeated execution of step S202-step S204 can be regarded as an iterative process. Specifically, the convergence of the function value after multiple optimizations of the objective function can be used as a condition for stopping iteration, but is not limited to this condition.
  • other conditions may also be used to stop iteration, for example, when iterating When the number of times is greater than the preset number, the iteration is stopped.
  • the optimized relative positional relationship between P 1 and P 2 , P 2 and P 3 , ..., P 19 and P 20 tends to converge, stop the iteration.
  • the relative positional relationship between P 1 and P 2 , P 2 and P 3 , ..., P 19 and P 20 after the optimization when the iteration is stopped can be used as the final optimized P 1 and P 2 , P 2 and P 3 , ..., the relative positional relationship between P 19 and P 20.
  • the difference between the point clouds detected by the detection device at different times can be determined The relative position of the target.
  • the target relative position relationship between the point clouds detected by the detection device at different times includes: the target relative position relationship between the point clouds that are not adjacent in time sequence detected by the detection device.
  • the target relative between non-adjacent point clouds in time sequence can be determined
  • the positional relationship for example, according to the relative positional relationship between P 1 and P 2 after the final optimization and the relative position relationship between P 2 and P 3 after the final optimization, it is possible to determine the P 1 and P 1 that are not adjacent in time sequence.
  • the relative positional relationship between P 3 and the target is possible to determine the P 1 and P 1 that are not adjacent in time sequence.
  • the relative positional relationship between adjacent point clouds in time sequence obtained by the detection device is used to aggregate the point clouds adjacent in time sequence to obtain multiple large point clouds, that is, multiple aggregate point clouds.
  • the detailed features in the point cloud are more abundant. Even if the adjacent area is blocked by the occluder, the matching calculation between the large point clouds can be completed through the detailed features in the large point cloud, so as to according to the large point cloud adjacent to the location
  • the relative position relationship between the two can determine a more accurate relative position relationship of the target between the point clouds detected by the detection device at different times, thereby improving the accuracy of the relative position relationship between the point clouds at different times.
  • FIG. 4 is a flowchart of a point cloud processing method provided by another embodiment of the application.
  • the relative position of the target between non-adjacent point clouds in the time series can be determined relationship. Therefore, an electronic map can be established according to the relative position relationship between adjacent point clouds in the final optimized time series and the target relative position relationship between non-adjacent point clouds in the time series. For example, according to the relative positional relationship between P 1 and P 2 , P 2 and P 3 , ..., P 19 and P 20 and other point clouds adjacent to each other in time series as shown in FIG. 3 , and P 1 and P 3
  • the relative positional relationship of the target between the point clouds that are not adjacent in time sequence can establish an electronic map of the road section 13 and the road section 14 as shown in FIG. 1.
  • the determining the relative positional relationship of the target between the point clouds detected by the detection device at different times according to the relative positional relationship between the aggregated point clouds adjacent to the position Can include:
  • Step S401 Optimize the relative position relationship between the aggregated point clouds that are adjacent to each other and the relative position relationship between the adjacent point clouds in the time series.
  • the objective function F is related to P 1 , P 2 , P 3 , P 4 , P 5 , P 6 , P 7 , P 8 , P 9 , P 10 , P 11 , P 12 , P 13 , P 14 , P 15 , P 16 , P 17 , P 18 , P 19 , P 20 , the conversion matrix between adjacent point clouds in time series is related, and the objective function is For optimization, the relative positional relationship between adjacent point clouds in time series such as P 1 and P 2 , P 2 and P 3 , ..., P 19 and P 20 can be optimized.
  • M 1 is related to the conversion matrix between adjacent point clouds in time sequence among P 1 , P 2 , P 3 , P 4 , and P 5
  • M 2 is related to P 6 , P 7 , P 8 , P 9 , P 10
  • M 4 is related to the conversion matrix between adjacent point clouds in the time sequence of P 11 , P 12 , P 13 , P 14 , and P 15
  • M 3 is related to In P 16 , P 17 , P 18 , P 19 , and P 20
  • the conversion matrix between adjacent point clouds in time series is related. Therefore, according to the optimized P 1 and P 2 , P 2 and P 3 ,..., P 19 and P 20 , the relative positional relationship between adjacent point clouds in time series can determine any two groups after optimization.
  • the optimizing the relative positional relationship between the aggregated point clouds that are adjacent to each other and the relative positional relationship between the adjacent point clouds in the time series includes: according to the aggregated points that are adjacent to the location Determine the objective function between the relative positional relationship between the clouds and the aggregated point clouds adjacent to the location; according to the objective function, optimize the relative positional relationship between the aggregated point clouds adjacent to the location and the time sequence upper phase The relative positional relationship between adjacent point clouds.
  • the determining the objective function according to the relative position relationship between the aggregated point clouds that are adjacent to each other and the aggregated point clouds that are adjacent to each other includes: according to one of the aggregated point clouds that are adjacent to each other.
  • the relative positional relationship between the adjacent aggregate point clouds and the adjacent aggregate point clouds are determined to determine the conversion error between the aggregate point clouds adjacent to the location; the conversion error between the aggregate point clouds adjacent to the location is determined.
  • the objective function can be obtained Through the objective function
  • the optimization can realize the optimization of the relative positional relationship between adjacent point clouds in time series such as P 1 and P 2 , P 2 and P 3 , ..., P 19 and P 20, and further, according to The optimized P 1 and P 2 , P 2 and P 3 , ..., P 19 and P 20 and the relative positional relationship between adjacent point clouds in time series can determine the relative position between any two groups of point clouds Positional relationship.
  • optimizing the relative positional relationship between the aggregated point clouds adjacent to each other and the relative positional relationship between adjacent point clouds in the time series includes: determining the optimized The relative positional relationship between the aggregated point clouds that are adjacent to each other and the relative positional relationship between the adjacent point clouds in the time sequence after optimization, so that the value of the objective function is less than or equal to a preset threshold .
  • the minimum function value of can optimize the relative positional relationship between adjacent point clouds in time series such as P 1 and P 2 , P 2 and P 3 ,..., P 19 and P 20, that is In other words, the optimized P 1 and P 2 , P 2 and P 3 , ..., the relative positional relationship between P 19 and P 20 , and the optimized D 1,2 , D 2,4 , D 4,3 , D 3,1 can minimize the function value of the objective function F, or make the function value of the objective function F less than or equal to a preset threshold.
  • Step S402 According to the optimized relative position relationship between the aggregated point clouds that are adjacent to each other and the optimized relative position relationship between the adjacent point clouds in the time sequence, determine that the detection device is at different times The relative positional relationship of the target between the point clouds obtained by the detection.
  • the target relative between non-adjacent point clouds in time sequence can be determined
  • the positional relationship for example, according to the relative positional relationship between P 1 and P 2 after the final optimization and the relative position relationship between P 2 and P 3 after the final optimization, it is possible to determine the P 1 and P 1 that are not adjacent in time sequence.
  • the relative positional relationship between P 3 and the target is possible to determine the P 1 and P 1 that are not adjacent in time sequence.
  • the timing can be determined according to the optimized D 1,2 The target relative positional relationship between P 1 and P 6 that are not adjacent to each other.
  • the relative positional relationship between adjacent point clouds in time sequence obtained by the detection device is used to aggregate the point clouds adjacent in time sequence to obtain multiple large point clouds, that is, multiple aggregate point clouds.
  • the detailed features in the point cloud are more abundant. Even if the adjacent areas are blocked by the occluder, the matching calculation between the large point clouds can be completed through the detailed features in the large point cloud, and the matching calculation between the large point clouds can be completed according to the adjacent large point clouds.
  • the relative positional relationship between the two can determine a more accurate relative positional relationship of the target between the point clouds detected by the detection device at different times, for example, the relative positional relationship between two sets of point clouds that are not adjacent in time series.
  • the embodiment of the present application provides a point cloud processing method.
  • the acquiring the relative position relationship between adjacent point clouds in the time series may include: acquiring the optimized relative position relationship between the adjacent point clouds in the time series.
  • the performing aggregation processing on the point clouds detected by the detection device according to the relative position relationship between the adjacent point clouds in the time sequence includes: according to the optimized adjacent point clouds in the time sequence The relative positional relationship between the point clouds of, the point clouds detected by the detection device are aggregated.
  • step S201 described in the foregoing embodiment is specifically acquiring the relative position relationship between adjacent point clouds in time series.
  • the ICP algorithm may be used to calculate the relative positional relationship between adjacent point clouds in time series.
  • step S201 can obtain the optimized relative positional relationship between adjacent point clouds in the time series, that is, according to the optimized P 1 and P 2 , P 2 and P 3 , ... ...
  • the relative positional relationship between P 19 and P 20 repeats the steps S202 to S204 described above.
  • the non-adjacent point clouds in the time series can be further improved.
  • the accuracy of the relative positional relationship between the targets can be further improved.
  • FIG. 5 is a structural diagram of a point cloud processing system provided by an embodiment of the application.
  • the point cloud processing system 50 includes: a detection device 51, a memory 52, and a processor 53.
  • the detection device 51 is used to detect and obtain a point cloud.
  • the point cloud obtained by the detection device 51 is shown in FIG. 6.
  • the white highlighted part may be the point cloud detected by the detection device 51.
  • the processor 53 may specifically be a component in the in-vehicle equipment in the foregoing embodiment, or other components, devices, or components equipped with a data processing function in the vehicle.
  • the memory 52 is used to store program code; the processor 53 calls the program code, and when the program code is executed, it is used to perform the following operations: obtain the relative positional relationship between adjacent point clouds in time series; For the relative positional relationship between adjacent point clouds in the time series, the point clouds detected by the detection device 51 are aggregated to obtain a plurality of aggregate point clouds, each of the aggregate point clouds includes at least two of the time sequences Adjacent point clouds on the upper side; determine the relative positional relationship between the adjacent aggregated point clouds among the multiple aggregated point clouds; determine the detection device according to the relative positional relationship between the adjacent aggregated point clouds 51. The relative position relationship of the target between the point clouds obtained by detection at different times.
  • the processor 53 determines the target relative position relationship between the point clouds detected by the detection device 51 at different times according to the relative position relationship between the aggregated point clouds at adjacent positions, it is specifically used for: optimization The relative positional relationship between the adjacent aggregate point clouds and the relative positional relationship between the adjacent point clouds in the time series; according to the optimized relative position between the adjacent aggregate point clouds The relationship and the optimized relative position relationship between adjacent point clouds in the time sequence determine the target relative position relationship between the point clouds detected by the detection device 51 at different times.
  • the processor 53 when the processor 53 optimizes the relative positional relationship between the aggregated point clouds that are adjacent to each other and the relative positional relationship between the adjacent point clouds in the time series, the processor 53 is specifically configured to: The relative positional relationship between adjacent aggregated point clouds and the adjacent aggregated point cloud are determined to determine the objective function; according to the objective function, the relative positional relationship between adjacent aggregated point clouds and the aggregated point cloud are optimized. Describe the relative positional relationship between adjacent point clouds in time series.
  • the processor 53 determines the objective function according to the relative positional relationship between the aggregated point clouds that are adjacent to each other and the aggregated point clouds that are adjacent to each other, the processor 53 is specifically configured to: The relative positional relationship between the aggregated point clouds and the aggregated point clouds at the adjacent positions are determined to determine the conversion error between the aggregated point clouds at adjacent positions; the conversion between the aggregated point clouds at adjacent positions is determined Error, determine the objective function.
  • the processor 53 when the processor 53 optimizes the relative positional relationship between the aggregated point clouds adjacent to each other and the relative positional relationship between adjacent point clouds in the time series according to the objective function, it is specifically configured to : Determine the optimized relative positional relationship between the aggregated point clouds in adjacent positions and the optimized relative positional relationship between adjacent point clouds in the time series, so that the value of the objective function is less than or Equal to the preset threshold.
  • the processor 53 obtains the relative positional relationship between adjacent point clouds in time series, it is specifically configured to: obtain the optimized relative positional relationship between adjacent point clouds in the time series; the processor 53 According to the relative positional relationship between adjacent point clouds in the time series, when the point clouds detected by the detection device 51 are aggregated, it is specifically used to: according to the optimized point clouds in the time series The relative positional relationship between the two is to perform aggregation processing on the point cloud detected by the detection device 51.
  • the processor 53 acquires the relative positional relationship between adjacent point clouds in time series, it is specifically configured to calculate the relative positional relationship between adjacent point clouds in time series according to the iterative closest point algorithm.
  • the processor 53 aggregates the point clouds detected by the detection device 51 according to the relative positional relationship between the adjacent point clouds in the time series, it is specifically configured to: The relative position relationship between the point clouds, the point clouds within the preset distance range are aggregated.
  • the processor 53 aggregates point clouds within a preset distance range according to the relative positional relationship between adjacent point clouds in the time series, it is specifically configured to:
  • the relative position relationship between the point clouds transforms the point clouds within the preset distance range into the same coordinate system.
  • the processor 53 aggregates the point clouds detected by the detection device 51 according to the relative position relationship between the adjacent point clouds in the time series, it is specifically configured to: The relative positional relationship between the point clouds, the point clouds within the preset time range are aggregated.
  • the processor 53 aggregates the point clouds within a preset time range according to the relative positional relationship between adjacent point clouds in the time series, it is specifically configured to:
  • the relative position relationship between the point clouds transforms the point clouds within the preset time range into the same coordinate system.
  • the target relative position relationship between the point clouds detected by the detection device 51 at different times includes: the target relative position relationship between the point clouds that are not adjacent in time sequence detected by the detection device 51.
  • the point cloud processing system provided by the embodiment of the present application can implement the above-mentioned point cloud processing method, and the specific principle and implementation manner of the point cloud processing method are similar to the above-mentioned embodiment, and will not be repeated here.
  • the embodiment of the application provides a movable platform.
  • the movable platform includes: a fuselage, a power system, and the point cloud processing system described in the above embodiment.
  • the power system is installed on the fuselage to provide moving power.
  • the point cloud processing system can implement the point cloud processing method as described above, and the specific principles and implementation manners of the point cloud processing method are similar to the foregoing embodiment, and will not be repeated here.
  • This embodiment does not limit the specific form of the movable platform.
  • the movable platform may be a drone, a movable robot, or a vehicle.
  • this embodiment also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the point cloud processing method described in the foregoing embodiment.
  • the disclosed device and method can be implemented in other ways.
  • the device embodiments described above are merely illustrative, for example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components may be combined or It can be integrated into another system, or some features can be ignored or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional units.
  • the above-mentioned integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium.
  • the above-mentioned software functional unit is stored in a storage medium, and includes several instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute the method described in each embodiment of the present application. Part of the steps.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Geometry (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Remote Sensing (AREA)
  • Computer Graphics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Electromagnetism (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Optical Radar Systems And Details Thereof (AREA)

Abstract

一种点云处理方法、***、可移动平台及存储介质，通过获得的时序上相邻的点云之间的相对位置关系，对时序上相邻的点云进行聚合处理，得到多个大点云即多个聚合点云，由于大点云中的细节特征更加丰富，即使位置相邻的区域被遮挡物所遮挡，也能通过大点云中的细节特征完成大点云之间的匹配计算，从而根据位置相邻的大点云之间的相对位置关系，可确定出较为精准的由探测设备在不同时刻探测获得的点云之间的目标相对位置关系，从而提高了不同时刻的点云之间的相对位置关系的精准度。

Description

点云处理方法、***、可移动平台及存储介质 技术领域
本申请实施例涉及探测技术领域，尤其涉及一种点云处理方法、***、可移动平台及存储介质。
背景技术
现有技术中在建立电子地图时，需要通过探测设备采集大量的点云，例如，探测设备可以在不同时刻采集不同的点云。进一步计算每个时刻采集的点云与其他时刻采集的点云之间的相对位置关系，并根据每个时刻的点云与其他时刻的点云之间的相对位置关系建立电子地图。
通常情况下，可以采用时序上相邻的点云之间的相对位置关系来求解时序上不相邻的点云之间的相对位置关系。但是，如果时序上不相邻的两组点云之间的时间跨度比较长，则会导致误差累积，从而导致不同时刻的点云之间的相对位置关系不精准。
发明内容
本申请实施例提供一种点云处理方法、***、可移动平台及存储介质，以提高探测设备在不同时刻探测获得的点云之间的相对位置关系的精准度。
本申请实施例的第一方面是提供一种点云处理方法，应用于可移动平台，所述可移动平台设置有探测设备，所述探测设备用于探测获得点云，所述方法包括：
获取时序上相邻的点云之间的相对位置关系；
根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，得到多个聚合点云，每个所述聚合点云包括至少两个所述时序上相邻的点云；
确定所述多个聚合点云中位置相邻的聚合点云之间的相对位置关系；
根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设 备在不同时刻探测获得的点云之间的目标相对位置关系。
本申请实施例的第二方面是提供一种点云处理***，包括：探测设备、存储器和处理器；
其中，所述探测设备用于探测获得点云；
所述存储器用于存储程序代码；
所述处理器，调用所述程序代码，当程序代码被执行时，用于执行以下操作：
获取时序上相邻的点云之间的相对位置关系；
根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，得到多个聚合点云，每个所述聚合点云包括至少两个所述时序上相邻的点云；
确定所述多个聚合点云中位置相邻的聚合点云之间的相对位置关系；
根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
本申请实施例的第三方面是提供一种可移动平台，包括：
机身；
动力***，安装在所述机身，用于提供移动动力；
以及第二方面所述的点云处理***。
本申请实施例的第四方面是提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行以实现第一方面所述的方法。
本实施例提供的点云处理方法、***、可移动平台及存储介质，通过获得的时序上相邻的点云之间的相对位置关系，对时序上相邻的点云进行聚合处理，得到多个大点云即多个聚合点云，由于大点云中的细节特征更加丰富，即使位置相邻的区域被遮挡物所遮挡，也能通过大点云中的细节特征完成大点云之间的匹配计算，从而根据位置相邻的大点云之间的相对位置关系，可确定出较为精准的由探测设备在不同时刻探测获得的点云之间的目标相对位置关系，从而提高了不同时刻的点云之间的相对位置关系的精准度。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的一种应用场景的示意图；
图2为本申请实施例提供的点云处理方法的流程图；
图3为本申请实施例提供的一种点云聚合的示意图；
图4为本申请另一实施例提供的点云处理方法的流程图；
图5为本申请实施例提供的点云处理***的结构图；
图6为本申请实施例提供的点云的示意图。
附图标记：
11：车辆；      12：服务器；           13：路段；
14：路段；      50：点云处理***；     51：探测设备；
52：存储器；    53：处理器。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
需要说明的是，当组件被称为“固定于”另一个组件，它可以直接在另一个组件上或者也可以存在居中的组件。当一个组件被认为是“连接”另一个组件，它可以是直接连接到另一个组件或者可能同时存在居中组件。
除非另有定义，本文所使用的所有的技术和科学术语与属于本申请的技术领域的技术人员通常理解的含义相同。本文中在本申请的说明书中所使用的术语只是为了描述具体的实施例的目的，不是旨在于限制本申请。本文所使用的术语“及/或”包括一个或多个相关的所列项目的任意的和所有的组合。
下面结合附图，对本申请的一些实施方式作详细说明。在不冲突的情况下，下述的实施例及实施例中的特征可以相互组合。
本申请实施例提供一种点云处理方法。该方法可应用于可移动平台，所述可移动平台设置有探测设备，该探测设备用于探测获得点云。在本实施例中，该可移动平台可以是无人机、可移动机器人或车辆。
本申请实施例以可移动平台是车辆为例，该车辆可以是无人驾驶车辆，或者是搭载有高级辅助驾驶(Advanced Driver Assistance Systems,ADAS)***的车辆等。如图1所示，车辆11为搭载有探测设备的载体，该探测设备具体可以是双目立体相机、飞行时间测距法(Time of flight，TOF)相机、毫米波雷达和/或激光雷达。车辆11在行驶的过程中，探测设备实时探测车辆11周围物体得到点云，在一些场景中点云也可称为三维点云。以激光雷达为例，当该激光雷达发射出的一束激光照射到物体表面时，该物体表面将会对该束激光进行反射，该激光雷达根据该物体表面反射的激光，可确定该物体相对于该激光雷达的方位、距离等信息。若该激光雷达发射出的该束激光按照某种轨迹进行扫描，例如360度旋转扫描，将得到大量的激光点，因而就可形成该物体的激光点云数据，也就是点云。
另外，本实施例并不限定点云处理方法的执行主体，该点云处理方法可以由车辆中的车载设备执行，也可以由车载设备之外的其他具有数据处理功能的设备执行，例如，如图1所示的服务器12，车辆11和服务器12可进行无线通信或有线通信，车辆11可以将探测设备探测获得的点云发送给服务器12，由服务器12执行该点云处理方法。下面以车载设备为例对本申请实施例提供的点云处理方法进行介绍。其中，车载设备可以是集成在车辆中控台中的具有数据处理功能的设备，或者也可以是放置在车辆内的平板电脑、手机、笔记本电脑等。
图2为本申请实施例提供的点云处理方法的流程图。如图2所示，本实施例中的方法，可以包括：
步骤S201、获取时序上相邻的点云之间的相对位置关系。
如图1所示，车辆11在行驶过程中，车辆11上搭载的探测设备实时探测车辆11周围物体得到点云，该探测设备可以和该车辆11上的车载设备通信连接，从而使得该车辆11上的车载设备可以实时获取到该探测设备探测得到的点云。例如，在t1时刻，车载设备获取到的点云记为P 1；在t2时刻，车载设备获取到的点云记为P 2；以此类推，在ti时刻，车载设备 获取到的点云记为P i。可以理解的是，在不同时刻，车载设备获取到的点云可包括多个三维点，每个三维点都对应有三维坐标。其中，t1、t2、…、ti可以是连续的时间点。因此，P 1和P 2、P 2和P 3、…、P i-1和P i可以看作是时序上相邻的点云。可以理解的是，相比于时序上不相邻的点云，时序上相邻的点云之间的重叠区域比较大，因此，可以根据时序上相邻的点云之间的重叠区域来确定时序上不相邻的点云之间的相对位置关系，该相对位置关系具体可包括旋转关系和/或平移关系。例如，P 1和P 2之间的相对位置关系包括P 1和P 2之间旋转关系和平移关系。
可选的，所述获取时序上相邻的点云之间的相对位置关系，包括：根据迭代最近点(IteratedClosestPoints,ICP)算法，计算所述时序上相邻的点云之间的相对位置关系。
例如，以P 1和P 2为例，p(i)为P 1中的任意一点，在P 2中确定出距离p(i)最近的一点m(i)，p(i)和m(i)可构成一组对应点对集，同理可以确定出P 1和P 2中的多组对应点对集，例如，p(1)和m(1)、p(2)和m(2)、……、p(n)和m(n)为n组对应点对集。其中，p(1)、p(2)、……、p(n)分别是点云P 1中的点，m(1)、m(2)、……、m(n)分别是点云P 2中的点，m(1)是点云P 2中距离p(1)最近的一个点，m(2)是点云P 2中距离p(2)最近的一个点，以此类推。
另外，需要说明的是，在本申请实施例中的大写字母和小写字母分别表示不同的含义，例如，P 1表示点云，p(1)表示点云P 1中的一个点。
由于P 1和P 2之间存在一定的旋转关系和平移关系，因此，每组对应点对集中的两个点之间也存在着一定的旋转关系和平移关系。例如，将P 1和P 2之间存在的旋转关系记为R，将P 1和P 2之间存在的平移关系记为T。n组对应点对集可建立n个方程组，进一步，运用数学方法对该n个方程组进行求解可得到P 1和P 2之间的旋转关系和平移关系。为了提高计算精度，还可以采用迭代算法，例如，ICP算法来计算P 1和P 2之间的旋转关系和平移关系。具体的迭代算法可有如下实现方式：
一种可能的实现方式是，将如上所述的n组对应点对集，例如，p(1)和m(1)、p(2)和m(2)、……、p(n)和m(n)代入公式(1)所述的函数E中：
Figure PCTCN2019104366-appb-000001
对公式(1)所述的函数E进行最优化求解，得到使得函数E的取值最 小或小于某一阈值的旋转关系R和平移关系T。
可以理解的是，其他的时序上相邻的点云之间的相对位置关系的求解过程可类似于P 1和P 2之间的相对位置关系的求解过程，此处不再一一赘述。
步骤S202、根据所述时序上相邻的点云之间的相对位置关系，对探测设备探测获得的点云进行聚合处理，得到多个聚合点云，每个所述聚合点云包括至少两个所述时序上相邻的点云。
如图1所示，假设车辆11沿着路段13行驶后调整方向继续沿着路段14行驶，路段13和路段14为来回路段。路段13和路段14之间可以通过护栏、绿化带等进行边界划分。在车辆11沿着来回路段行驶的过程中，假设探测设备探测获得的点云依次为P 1、P 2、……、P 20，例如图3所示。
根据如上所述的迭代最近点算法可计算出P 1、P 2、……、P 20中每两个时序上相邻的点云之间的相对位置关系，即计算出P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系，其中，将P 1和P 2之间的旋转关系记为R P1,P2，将P 1和P 2之间的平移关系记为T P1,P2，将P 2和P 3之间的旋转关系记为R P2,P3，将P 2和P 3之间的平移关系记为T P2,P3，以此类推，将P 19和P 20之间的旋转关系记为R P19,P20，将P 19和P 20之间的平移关系记为T P19,P20。
另外，根据该每两个时序上相邻的点云之间的相对位置关系，还可以对P 1、P 2、……、P 20进行聚合处理，例如，将P 1、P 2、P 3、P 4、P 5聚合为一个聚合点云记为M 1，将P 6、P 7、P 8、P 9、P 10聚合为一个聚合点云记为M 2，将P 11、P 12、P 13、P 14、P 15聚合为一个聚合点云记为M 4，将P 16、P 17、P 18、P 19、P 20聚合为一个聚合点云记为M 3。可以理解的是，每个聚合点云中包括的点云的数量可以相同，也可以不同，并且本实施例也不限定每个聚合点云可包括的点云的数量，可选的，每个聚合点云至少包括两个时序上相邻的点云。
具体的，对探测设备探测获得的多个点云进行聚合处理时可包括如下几种可能的情况：
在一种可能的实现方式中，所述根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，包括：根据 所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理。
例如，从P 1开始，按照时间的先后顺序，找到下一时刻的点云P 2，并计算P 2相对于P 1的距离，如果P 2相对于P 1的距离小于预设的距离阈值，例如100米，则继续寻找下一个点云P 3，如果P 3相对于P 1的距离小于预设的距离阈值，则继续寻找下一个点云，直到找到一个点云例如P 6，P 6相对于P 1的距离大于该预设的距离阈值，则将P 1、P 2、P 3、P 4、P 5聚合为一个聚合点云M 1。进一步，从P 6开始，按照时间的先后顺序，继续寻找下一个点云，并重复执行前面所述的聚合过程，得到P 6、P 7、P 8、P 9、P 10聚合成的聚合点云M 2，P 11、P 12、P 13、P 14、P 15聚合成的聚合点云M 4，P 16、P 17、P 18、P 19、P 20聚合成的聚合点云M 3。
其中，两个点云之间的距离具体可以是将两个点云投影到同一坐标系下计算得到的两个点云之间的距离。以P 2相对于P 1的距离为例，可以将P 2投影到P 1所在的坐标系中，例如，将P 2投影到P 1所在的坐标系中得到的投影点云为P' 2，在该坐标系中可以将P' 2相对于P 1的距离作为P 2相对于P 1的距离。其中，P' 2相对于P 1的距离可以有多种计算方式，例如，P' 2相对于P 1的距离可以是P' 2的中心点与P 1的中心点之间的距离，可以理解此处只是一个举例，在其他实施例中还可以有其他不同的计算方式。
可选的，所述根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理，包括：根据所述时序上相邻的点云之间的相对位置关系，将所述预设距离范围内的点云转换到同一个坐标系中。
例如，在将P 1、P 2、P 3、P 4、P 5聚合为M 1的过程中，可以将P 1、P 2、P 3、P 4、P 5转换到同一个坐标系中，例如，将P 2、P 3、P 4、P 5转换到P 1所在的坐标系中。同理，在将P 6、P 7、P 8、P 9、P 10聚合为M 2的过程中，可以将P 7、P 8、P 9、P 10转换到P 6所在的坐标系中。在将P 11、P 12、P 13、P 14、P 15聚合为M 4的过程中，可以将P 12、P 13、P 14、P 15转换到P 11所在的坐标系中。在将P 16、P 17、P 18、P 19、P 20聚合为M 3的过程中，可以将P 17、P 18、P 19、P 20转换到P 16所在的坐标系中。
在另一种可能的实现方式中，所述根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，包括：根 据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理。
例如，从P 1开始，P 1、P 2、P 3、P 4、P 5是在预设时间范围，例如，30秒内探测得到的点云，则可以将P 1、P 2、P 3、P 4、P 5聚合为一个聚合点云M 1。同理，从P 6开始，P 6、P 7、P 8、P 9、P 10是在预设时间范围探测得到的点云，则将P 6、P 7、P 8、P 9、P 10聚合为M 2。从P 11开始，P 11、P 12、P 13、P 14、P 15是在预设时间范围探测得到的点云，则将P 11、P 12、P 13、P 14、P 15聚合为M 4。从P 16开始，P 16、P 17、P 18、P 19、P 20是在预设时间范围探测得到的点云，则将P 16、P 17、P 18、P 19、P 20聚合为M 3。
可选的，所述根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理，包括：根据所述时序上相邻的点云之间的相对位置关系，将所述预设时间范围内的点云转换到同一个坐标系中。
例如，在将P 1、P 2、P 3、P 4、P 5聚合为M 1的过程中，可以将P 1、P 2、P 3、P 4、P 5转换到同一个坐标系中，例如，将P 2、P 3、P 4、P 5转换到P 1所在的坐标系中。此时，M 1可表示为M 1＝P 1+K 1,2P 2+K 1,3P 3+K 1,4P 4+K 1,5P 5，其中，K 1，2表示P 2和P 1之间的转换矩阵，K 1，2可以根据P 1和P 2之间的旋转关系R P1,P2、以及P 1和P 2之间的平移关系T P1,P2得到，在本实施例中，转换矩阵也就是如上所述的相对位置关系。K 1,3表示P 3和P 1之间的转换矩阵，K 1,3可根据K 1，2和K 2,3得到，其中，K 2,3表示P 3和P 2之间的转换矩阵，K 2,3可以根据P 2和P 3之间的旋转关系R P2,P3、以及P 2和P 3之间的平移关系T P2,P3得到。同理，K 1,4表示P 4和P 1之间的转换矩阵，K 1,4可根据K 1，2、K 2,3、K 3,4得到，K 3,4表示P 4和P 3之间的转换矩阵，K 3,4可以根据P 4和P 3之间的旋转关系和平移关系得到。K 1,5表示P 5和P 1之间的转换矩阵，K 1,5可根据K 1，2、K 2,3、K 3,4、K 4,5得到，K 4,5表示P 4和P 5之间的转换矩阵，K 4,5可以根据P 4和P 5之间的旋转关系和平移关系得到。也就是说，M 1与P 1、P 2、P 3、P 4、P 5中时序上相邻的点云之间的转换矩阵相关。
同理，在将P 6、P 7、P 8、P 9、P 10聚合为M 2的过程中，可以将P 7、P 8、P 9、P 10转换到P 6所在的坐标系中。此时，M 2可表示为M 2＝P 6+K 6,7P 7+K 6,8P 8+K 6,9P 9+K 6,10P 10，其中，K 6,7表示P 7和P 6之间的转换矩阵，K 6,8表示P 8和P 6之间的转换矩阵，K 6,9表示P 9和P 6之间的转换矩阵，K 6,10表示P 10和P 6之间的转换矩阵。K 6,7、K 6,8、K 6,9、K 6,10的计算过程类似于如上所述的K 1，2、K 1,3、K 1,4、K 1,5的计算过程，此处不再赘述。也就是说，M 2与P 6、P 7、P 8、P 9、P 10中时序上相邻的点云之间的转换矩阵相关。
同理，在将P 11、P 12、P 13、P 14、P 15聚合为M 4的过程中，可以将P 12、P 13、P 14、P 15转换到P 11所在的坐标系中。此时，M 4可表示为M 4＝P 11+K 11,12P 12+K 11,13P 13+K 11,14P 14+K 11,15P 15，其中，K 11,12、K 11,13、K 11,14、K 11,15的含义以及计算过程均与如上所述的K 1，2、K 1,3、K 1,4、K 1,5的含义及计算过程类似，此处不再赘述。也就是说，M 4与P 11、P 12、P 13、P 14、P 15中时序上相邻的点云之间的转换矩阵相关。
同理，在将P 16、P 17、P 18、P 19、P 20聚合为M 3的过程中，可以将P 17、P 18、P 19、P 20转换到P 16所在的坐标系中。此时，M 3可表示为M 3＝P 16+K 16,17P 17+K 16,18P 18+K 16,19P 19+K 16,20P 20。其中，K 16,17、K 16,18、K 16,19、K 16,20的含义以及计算过程均与如上所述的K 1，2、K 1,3、K 1,4、K 1,5的含义及计算过程类似，此处不再赘述。也就是说，M 3与P 16、P 17、P 18、P 19、P 20中时序上相邻的点云之间的转换矩阵相关。
步骤S203、确定所述多个聚合点云中位置相邻的聚合点云之间的相对位置关系。
在一种可能的实现方式中，在确定出多个聚合点云后，根据定位模块，例如全球定位***(Global Positioning System，GPS)，可以确定出该多个聚合点云中位置相邻的聚合点云，其中，聚合点云的GPS信息可以是 该聚合点云中的中间点云的GPS信息，中间点云的GPS信息具体可以是探测设备在探测获得该中间点云时车辆的GPS信息。例如，P 3是聚合点云M 1中的中间点云，则该探测设备在探测获得P 3时车辆的GPS信息可作为聚合点云M 1的GPS信息。同理，P 8是聚合点云M 2中的中间点云，则该探测设备在探测获得P 8时车辆的GPS信息可作为聚合点云M 2的GPS信息。P 13是聚合点云M 4中的中间点云，则该探测设备在探测获得P 13时车辆的GPS信息可作为聚合点云M 4的GPS信息。P 18是聚合点云M 3中的中间点云，则该探测设备在探测获得P 18时车辆的GPS信息可作为聚合点云M 3的GPS信息。
进一步，根据聚合点云M 1、M 2、M 3、M 4的GPS信息，确定出M 1、M 2、M 3、M 4中位置相邻的聚合点云。例如，根据聚合点云M 1、M 2的GPS信息，确定M 1和M 2之间的距离在预设距离范围内，例如，在10米以内，则将M 1和M 2作为位置相邻的聚合点云，同理，可确定M 2和M 4为位置相邻的聚合点云，M 3和M 4为位置相邻的聚合点云，M 3和M 1为位置相邻的聚合点云。
进一步，计算M 1和M 2之间的相对位置关系，即M 1和M 2之间的旋转关系和平移关系，以及计算M 2和M 4之间的相对位置关系、M 3和M 4之间的相对位置关系、M 3和M 1之间的相对位置关系。具体的，可采用如上所述的ICP算法计算位置相邻的聚合点云之间的相对位置关系，具体的计算过程可以参考如上所述的P 1和P 2之间的相对位置关系的计算过程，此处不再赘述。例如，M 1和M 2之间的旋转关系记为R M1,M2，M 1和M 2之间的平移关系记为T M1,M2；M 2和M 4之间的旋转关系记为R M2,M4，M 2和M 4之间的平移关系记为T M2,M4；M 3和M 4之间的旋转关系记为R M4,M3，M 3和M 4之间的平移关系记为T M4,M3；M 3和M 1之间的旋转关系记为R M3,M1，M 3和M 1之间的平移关系记为T M3,M1。
需要说明的是，上述定位模块并不限于GPS，还可以是北斗、伽利略 (Galileo)、格洛纳斯(GLONASS)等。
步骤S204、根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
例如，根据M 1和M 2之间的旋转关系R M1,M2、以及M 1和M 2之间的平移关系T M1,M2，可得到M 1和M 2之间的转换矩阵D 1,2，该转换矩阵D 1,2可以是M 2到M 1的转换矩阵。同理，根据M 2和M 4之间的旋转关系R M2,M4、以及M 2和M 4之间的平移关系T M2,M4，可得到M 2和M 4之间的转换矩阵D 2,4，该转换矩阵D 2,4可以是M 4到M 2的转换矩阵。根据M 3和M 4之间的旋转关系R M4,M3、以及M 3和M 4之间的平移关系T M4,M3，可得到M 3和M 4之间的转换矩阵D 4,3，该转换矩阵D 4,3可以是M 3到M 4的转换矩阵。根据M 3和M 1之间的旋转关系R M3,M1、以及M 3和M 1之间的平移关系T M3,M1，可得到M 3和M 1之间的转换矩阵D 3,1，该转换矩阵D 3,1可以是M 1到M 3的转换矩阵。在本实施例中，转换矩阵也就是如上所述的相对位置关系。
进一步，采用M 1和M 2之间的转换矩阵D 1,2对M 2进行转换，得到M' 1，M' 1＝D 1,2M 2。可以理解，M' 1和M 1可能不完全相同，也就是说，M' 1和M 1之间可能存在一定的误差，该误差可记为e 1,2，e 1,2＝(M 1-D 1,2M 2)。由于在将P 1、P 2、P 3、P 4、P 5聚合为M 1的过程中，可以将P 2、P 3、P 4、P 5转换到P 1所在的坐标系中，在将P 6、P 7、P 8、P 9、P 10聚合为M 2的过程中，可以将P 7、P 8、P 9、P 10转换到P 6所在的坐标系中。因此，M 1和M 2之间的转换矩阵D 1,2具体为P 6和P 1之间的转换矩阵K 1,6。由于M 1与P 1、P 2、P 3、P 4、P 5中时序上相邻的点云之间的转换矩阵相关，M 2与P 6、P 7、P 8、P 9、P 10中时序上相邻的点云之间的转换矩阵相关，因此，e 1,2与P 1、P 2、P 3、P 4、P 5、P 6、P 7、P 8、P 9、P 10中时序上相邻的点云之间的转换矩阵相关。
同理，采用M 2和M 4之间的转换矩阵D 2,4对M 4进行转换，得到M' 2，M' 2＝D 2,4M 4，M' 2和M 2之间可能存在的误差记为e 2,4，e 2,4＝(M 2-D 2,4M 4)。由于在将P 6、P 7、P 8、P 9、P 10聚合为M 2的过程中，可将P 7、P 8、P 9、P 10转换到P 6所在的坐标系中，在将P 11、P 12、P 13、P 14、P 15聚合为M 4的过程中，可以将P 12、P 13、P 14、P 15转换到P 11所在的坐标系中，因此，M 2和M 4之间的转换矩阵D 2,4具体为P 6和P 11之间的转换矩阵。另外，M 2与P 6、P 7、P 8、P 9、P 10中时序上相邻的点云之间的转换矩阵相关，M 4与P 11、P 12、P 13、P 14、P 15中时序上相邻的点云之间的转换矩阵相关，因此，e 2,4与P 6、P 7、P 8、P 9、P 10、P 11、P 12、P 13、P 14、P 15中时序上相邻的点云之间的转换矩阵相关。
同理，采用M 3和M 4之间的转换矩阵D 4,3对M 3进行转换，得到M' 4，M' 4＝D 4,3M 3，M 4和M' 4之间可能存在的误差记为e 4,3，e 4,3＝(M 4-D 4,3M 3)。e 4,3与P 11、P 12、P 13、P 14、P 15、P 16、P 17、P 18、P 19、P 20中时序上相邻的点云之间的转换矩阵相关。
同理，采用M 3和M 1之间的转换矩阵D 3,1对M 1进行转换，得到M' 3，M' 3＝D 3,1M 1，M 3和M' 3之间可能存在的误差记为e 3,1，e 3,1＝(M 3-D 3,1M 1)。e 3,1与P 1、P 2、P 3、P 4、P 5、P 6、P 7、P 8、P 9、P 10、P 11、P 12、P 13、P 14、P 15、P 16、P 17、P 18、P 19、P 20中时序上相邻的点云之间的转换矩阵相关。
根据如上所述的误差e 1,2、e 2,4、e 4,3和e 3,1可得到目标函数
Figure PCTCN2019104366-appb-000002
其中，目标函数F与P 1、P 2、P 3、P 4、P 5、P 6、P 7、P 8、P 9、P 10、P 11、P 12、P 13、P 14、P 15、P 16、P 17、P 18、P 19、P 20中时序上相邻的点云之间的转换矩阵相关。因此，通过对该目标函数
Figure PCTCN2019104366-appb-000003
进行优化，即求解该目标函数的最小函数值，可对如上所述的P 1和P 2、P 2和P 3、……、P 19和P 20等时 序上相邻的点云之间的相对位置关系进行优化，也就是说，优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系可使得该目标函数F的函数值最小，或者使得该目标函数F的函数值小于或等于预设阈值。
进一步，根据优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系重复执行如上所述的步骤S202-步骤S204，可以理解的是，本实施例并不限定重复执行的次数。从而使得再次优化或多次优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系更加精准，同时也使得每次优化后的该目标函数F的函数值逐渐收敛。另外，每重复执行一次步骤S202-步骤S204，可作为一次迭代过程。具体的，该目标函数多次优化后的函数值变得收敛可以作为停止迭代的一个条件，但不限于这一个条件，在其他实施例中，还可以通过其他条件来停止迭代，例如，当迭代的次数大于预设次数时，停止迭代。或者，优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系趋于收敛时，停止迭代。在本实施例中，可以将停止迭代时优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系作为最终优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系。
进一步，根据最终优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系，可确定出探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
可选的，所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系，包括：所述探测设备探测获得的时序上不相邻的点云之间的目标相对位置关系。
例如，根据最终优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系，可确定出时序上不相邻的点云之间的目标相对位置关系，例如，根据最终优化后的P 1和P 2之间的相对位置关系和最终优化后的P 2和P 3之间的相对位置关系，可确定出时序上不相邻的P 1和P 3之间的目标相对位置关系。
由于最终优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系是较为精准的，因此，根据最终优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系得到的时序上不相邻的点云之间的目标相对位 置关系也是精准的。
本实施例通过探测设备探测获得的时序上相邻的点云之间的相对位置关系，对时序上相邻的点云进行聚合处理，得到多个大点云即多个聚合点云，由于大点云中的细节特征更加丰富，即使位置相邻的区域被遮挡物所遮挡，也能通过大点云中的细节特征完成大点云之间的匹配计算，从而根据位置相邻的大点云之间的相对位置关系，可确定出较为精准的由探测设备在不同时刻探测获得的点云之间的目标相对位置关系，从而提高了不同时刻的点云之间的相对位置关系的精准度。
本申请实施例提供一种点云处理方法。图4为本申请另一实施例提供的点云处理方法的流程图。如图4所示，在上述实施例的基础上，根据最终优化后的时序上相邻的点云之间的相对位置关系，可确定出时序上不相邻的点云之间的目标相对位置关系。因此，根据最终优化后的时序上相邻的点云之间的相对位置关系、以及时序上不相邻的点云之间的目标相对位置关系，可建立电子地图。例如，根据如图3所示的P 1和P 2、P 2和P 3、……、P 19和P 20等时序上相邻的点云之间的相对位置关系、以及P 1和P 3、P 1和P 4等时序上不相邻的点云之间的目标相对位置关系，可建立如图1所示的路段13和路段14的电子地图。
另外，在上述实施例的基础上，所述根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系，可以包括：
步骤S401、优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系。
根据上述内容可知，目标函数F与P 1、P 2、P 3、P 4、P 5、P 6、P 7、P 8、P 9、P 10、P 11、P 12、P 13、P 14、P 15、P 16、P 17、P 18、P 19、P 20中时序上相邻的点云之间的转换矩阵相关，通过对该目标函数
Figure PCTCN2019104366-appb-000004
进行优化，可对如上所述的P 1和P 2、P 2和P 3、……、P 19和P 20等时序上相邻的点云之间的相对位置关系进行优化。由于M 1与P 1、P 2、P 3、P 4、P 5中时序上相邻的点 云之间的转换矩阵相关，M 2与P 6、P 7、P 8、P 9、P 10中时序上相邻的点云之间的转换矩阵相关，M 4与P 11、P 12、P 13、P 14、P 15中时序上相邻的点云之间的转换矩阵相关，M 3与P 16、P 17、P 18、P 19、P 20中时序上相邻的点云之间的转换矩阵相关。因此，根据优化后的P 1和P 2、P 2和P 3、……、P 19和P 20等时序上相邻的点云之间的相对位置关系，可确定出优化后的任意两组点云之间的相对位置关系。
可选的，所述优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系，包括：根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数；根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系。
可选的，所述根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数，包括：根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定所述位置相邻的聚合点云之间的转换误差；根据所述位置相邻的聚合点云之间的转换误差，确定所述目标函数。
例如，根据如上所述的误差e 1,2、e 2,4、e 4,3和e 3,1可得到目标函数
Figure PCTCN2019104366-appb-000005
通过对该目标函数
Figure PCTCN2019104366-appb-000006
进行优化，可以实现对如上所述的P 1和P 2、P 2和P 3、……、P 19和P 20等时序上相邻的点云之间的相对位置关系的优化，进一步，根据优化后的P 1和P 2、P 2和P 3、……、P 19和P 20等时序上相邻的点云之间的相对位置关系，可确定出任意两组点云之间的相对位置关系。
可选的，所述根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系，包括：确定优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，以使所述目标函数的值小于或 等于预设阈值。
例如，通过求解目标函数
Figure PCTCN2019104366-appb-000007
的最小函数值，可对如上所述的P 1和P 2、P 2和P 3、……、P 19和P 20等时序上相邻的点云之间的相对位置关系进行优化，也就是说，优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系、以及优化后的D 1,2、D 2,4、D 4,3、D 3,1可使得该目标函数F的函数值最小，或者使得该目标函数F的函数值小于或等于预设阈值。
步骤S402、根据优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
例如，根据最终优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系，可确定出时序上不相邻的点云之间的目标相对位置关系，例如，根据最终优化后的P 1和P 2之间的相对位置关系和最终优化后的P 2和P 3之间的相对位置关系，可确定出时序上不相邻的P 1和P 3之间的目标相对位置关系。
再例如，由于M 1和M 2之间的转换矩阵D 1,2具体为P 6和P 1之间的转换矩阵K 1,6，因此，根据优化后的D 1,2，可确定出时序上不相邻的P 1和P 6之间的目标相对位置关系。
本实施例通过探测设备探测获得的时序上相邻的点云之间的相对位置关系，对时序上相邻的点云进行聚合处理，得到多个大点云即多个聚合点云，由于大点云中的细节特征更加丰富，即使位置相邻的区域通过遮挡物遮挡，也能通过大点云中的细节特征完成大点云之间的匹配计算，从而根据位置相邻的大点云之间的相对位置关系，可确定出较为精准的由探测设备在不同时刻探测获得的点云之间的目标相对位置关系，例如，时序上不相邻的两组点云之间的相对位置关系。因此，相比于现有技术中采用时序上相邻的点云之间的相对位置关系来求解时序上不相邻的两组点云之间的相对位置关系，可避免时序上不相邻的两组点云之间的时间跨度比较长而导致的误差累积，从而提高了时序上不相邻的两组点云之间的相对位置关系的精度。另外，根据最终优化后的时序上相邻的点云之间的相对位 置关系、以及时序上不相邻的点云之间的目标相对位置关系，可建立出精度较高的电子地图。
本申请实施例提供一种点云处理方法。在上述实施例的基础上，所述获取时序上相邻的点云之间的相对位置关系，可以包括：获取优化后的所述时序上相邻的点云之间的相对位置关系。
相应的，所述根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，包括：根据所述优化后的所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理。
例如，上述实施例所述的步骤S201具体为获取时序上相邻的点云之间的相对位置关系。在上述实施例中，可采用ICP算法来计算时序上相邻的点云之间的相对位置关系。在本实施例中，步骤S201可以获取优化后的所述时序上相邻的点云之间的相对位置关系，也就是说，根据优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系重复执行如上所述的步骤S202-步骤S204。例如，根据优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系，重新对P 1、P 2、……、P 20进行聚合处理，得到多个聚合点云，此处的多个聚合点云可能与如上所述的M 1、M 2、M 3和M 4不同，也有可能相同。进一步，确定出重新聚合得到的多个聚合点云中位置相邻的聚合点云，并根据该位置相邻的聚合点云之间的相对位置关系、以及该位置相邻的聚合点云，确定出如上所述的目标函数F，再次对该目标函数F进行优化，得到再次优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系。再次优化后的P 1和P 2、P 2和P 3、……、P 19和P 20之间的相对位置关系可以进一步作为步骤S201中获取的优化后的所述时序上相邻的点云之间的相对位置关系，从而不断的重复执行S201-S204。
本实施例通过获取优化后的所述时序上相邻的点云之间的相对位置关系，并根据优化后的时序上相邻的点云之间的相对位置关系，重新进行点云聚合，得到多个聚合点云，进一步，确定出重新聚合得到的多个聚合点云中位置相邻的聚合点云，并根据该位置相邻的聚合点云之间的相对位置关系、以及该位置相邻的聚合点云，确定出目标函数，再次对该目标函 数进行优化，得到再次优化后的时序上相邻的点云之间的相对位置关系，通过不断的重复迭代，可进一步提高时序上相邻的点云之间的相对位置关系的精度。当根据精度较高的时序上相邻的点云之间的相对位置关系来确定时序上不相邻的点云之间的目标相对位置关系时，还可以进一步提高时序上不相邻的点云之间的目标相对位置关系的精度。
本申请实施例提供一种点云处理***。图5为本申请实施例提供的点云处理***的结构图，如图5所示，点云处理***50包括：探测设备51、存储器52和处理器53。其中，探测设备51用于探测获得点云，探测设备51探测获得的点云例如图6所示，在图6中，白色高亮的部分可以是探测设备51探测获得的点云。另外，处理器53具体可以是上述实施例中车载设备中的部件，或者是车辆中搭载的具有数据处理功能的其他部件、器件或组件。具体的，存储器52用于存储程序代码；处理器53，调用所述程序代码，当程序代码被执行时，用于执行以下操作：获取时序上相邻的点云之间的相对位置关系；根据所述时序上相邻的点云之间的相对位置关系，对探测设备51探测获得的点云进行聚合处理，得到多个聚合点云，每个所述聚合点云包括至少两个所述时序上相邻的点云；确定所述多个聚合点云中位置相邻的聚合点云之间的相对位置关系；根据所述位置相邻的聚合点云之间的相对位置关系，确定探测设备51在不同时刻探测获得的点云之间的目标相对位置关系。
可选的，处理器53根据所述位置相邻的聚合点云之间的相对位置关系，确定探测设备51在不同时刻探测获得的点云之间的目标相对位置关系时，具体用于：优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系；根据优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，确定探测设备51在不同时刻探测获得的点云之间的目标相对位置关系。
可选的，处理器53优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系时，具体用于：根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确 定目标函数；根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系。
可选的，处理器53根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数时，具体用于：根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定所述位置相邻的聚合点云之间的转换误差；根据所述位置相邻的聚合点云之间的转换误差，确定所述目标函数。
可选的，处理器53根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系时，具体用于：确定优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，以使所述目标函数的值小于或等于预设阈值。
可选的，处理器53获取时序上相邻的点云之间的相对位置关系时，具体用于：获取优化后的所述时序上相邻的点云之间的相对位置关系；处理器53根据所述时序上相邻的点云之间的相对位置关系，对探测设备51探测获得的点云进行聚合处理时，具体用于：根据所述优化后的所述时序上相邻的点云之间的相对位置关系，对探测设备51探测获得的点云进行聚合处理。
可选的，处理器53获取时序上相邻的点云之间的相对位置关系时，具体用于：根据迭代最近点算法，计算所述时序上相邻的点云之间的相对位置关系。
可选的，处理器53根据所述时序上相邻的点云之间的相对位置关系，对探测设备51探测获得的点云进行聚合处理时，具体用于：根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理。
可选的，处理器53根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理时，具体用于：根据所述时序上相邻的点云之间的相对位置关系，将所述预设距离范围内的点云转换到同一个坐标系中。
可选的，处理器53根据所述时序上相邻的点云之间的相对位置关系， 对探测设备51探测获得的点云进行聚合处理时，具体用于：根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理。
可选的，处理器53根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理时，具体用于：根据所述时序上相邻的点云之间的相对位置关系，将所述预设时间范围内的点云转换到同一个坐标系中。
可选的，探测设备51在不同时刻探测获得的点云之间的目标相对位置关系，包括：探测设备51探测获得的时序上不相邻的点云之间的目标相对位置关系。
本申请实施例提供的点云处理***可以实现如上所述的点云处理方法，该点云处理方法的具体原理和实现方式均与上述实施例类似，此处不再赘述。
本申请实施例提供一种可移动平台。该可移动平台包括：机身、动力***和如上实施例所述的点云处理***。其中，动力***安装在所述机身，用于提供移动动力。点云处理***可以实现如上所述的点云处理方法，该点云处理方法的具体原理和实现方式均与上述实施例类似，此处不再赘述。本实施例并不限定该可移动平台的具体形态，例如，该可移动平台可以是无人机、可移动机器人或车辆等。
另外，本实施例还提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行以实现上述实施例所述的点云处理方法。
在本申请所提供的几个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个***，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接 耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本申请各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元，可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本申请各个实施例所述方法的部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
本领域技术人员可以清楚地了解到，为描述的方便和简洁，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将装置的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。上述描述的装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
最后应说明的是：以上各实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述各实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的范围。

Claims (26)

  1. 一种点云处理方法，其特征在于，应用于可移动平台，所述可移动平台设置有探测设备，所述探测设备用于探测获得点云，所述方法包括：
    获取时序上相邻的点云之间的相对位置关系；
    根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，得到多个聚合点云，每个所述聚合点云包括至少两个所述时序上相邻的点云；
    确定所述多个聚合点云中位置相邻的聚合点云之间的相对位置关系；
    根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
  2. 根据权利要求1所述的方法，其特征在于，所述根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系，包括：
    优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系；
    根据优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
  3. 根据权利要求2所述的方法，其特征在于，所述优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系，包括：
    根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数；
    根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系。
  4. 根据权利要求3所述的方法，其特征在于，所述根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数，包括：
    根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定所述位置相邻的聚合点云之间的转换误差；
    根据所述位置相邻的聚合点云之间的转换误差，确定所述目标函数。
  5. 根据权利要求3或4所述的方法，其特征在于，所述根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系，包括：
    确定优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，以使所述目标函数的值小于或等于预设阈值。
  6. 根据权利要求5所述的方法，其特征在于，所述获取时序上相邻的点云之间的相对位置关系，包括：
    获取优化后的所述时序上相邻的点云之间的相对位置关系；
    所述根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，包括：
    根据所述优化后的所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理。
  7. 根据权利要求1所述的方法，其特征在于，所述获取时序上相邻的点云之间的相对位置关系，包括：
    根据迭代最近点算法，计算所述时序上相邻的点云之间的相对位置关系。
  8. 根据权利要求1所述的方法，其特征在于，所述根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，包括：
    根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理。
  9. 根据权利要求8所述的方法，其特征在于，所述根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理，包括：
    根据所述时序上相邻的点云之间的相对位置关系，将所述预设距离范围内的点云转换到同一个坐标系中。
  10. 根据权利要求1所述的方法，其特征在于，所述根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚 合处理，包括：
    根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理。
  11. 根据权利要求10所述的方法，其特征在于，所述根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理，包括：
    根据所述时序上相邻的点云之间的相对位置关系，将所述预设时间范围内的点云转换到同一个坐标系中。
  12. 根据权利要求1-11任一项所述的方法，其特征在于，所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系，包括：
    所述探测设备探测获得的时序上不相邻的点云之间的目标相对位置关系。
  13. 一种点云处理***，其特征在于，包括：探测设备、存储器和处理器；
    其中，所述探测设备用于探测获得点云；
    所述存储器用于存储程序代码；
    所述处理器，调用所述程序代码，当程序代码被执行时，用于执行以下操作：
    获取时序上相邻的点云之间的相对位置关系；
    根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理，得到多个聚合点云，每个所述聚合点云包括至少两个所述时序上相邻的点云；
    确定所述多个聚合点云中位置相邻的聚合点云之间的相对位置关系；
    根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
  14. 根据权利要求13所述的***，其特征在于，所述处理器根据所述位置相邻的聚合点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系时，具体用于：
    优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系；
    根据优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，确定所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系。
  15. 根据权利要求14所述的***，其特征在于，所述处理器优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系时，具体用于：
    根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数；
    根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系。
  16. 根据权利要求15所述的***，其特征在于，所述处理器根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定目标函数时，具体用于：
    根据所述位置相邻的聚合点云之间的相对位置关系和所述位置相邻的聚合点云，确定所述位置相邻的聚合点云之间的转换误差；
    根据所述位置相邻的聚合点云之间的转换误差，确定所述目标函数。
  17. 根据权利要求15或16所述的***，其特征在于，所述处理器根据所述目标函数，优化所述位置相邻的聚合点云之间的相对位置关系和所述时序上相邻的点云之间的相对位置关系时，具体用于：
    确定优化后的所述位置相邻的聚合点云之间的相对位置关系和优化后的所述时序上相邻的点云之间的相对位置关系，以使所述目标函数的值小于或等于预设阈值。
  18. 根据权利要求17所述的***，其特征在于，所述处理器获取时序上相邻的点云之间的相对位置关系时，具体用于：
    获取优化后的所述时序上相邻的点云之间的相对位置关系；
    所述处理器根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理时，具体用于：
    根据所述优化后的所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理。
  19. 根据权利要求13所述的***，其特征在于，所述处理器获取时 序上相邻的点云之间的相对位置关系时，具体用于：
    根据迭代最近点算法，计算所述时序上相邻的点云之间的相对位置关系。
  20. 根据权利要求13所述的***，其特征在于，所述处理器根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理时，具体用于：
    根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理。
  21. 根据权利要求20所述的***，其特征在于，所述处理器根据所述时序上相邻的点云之间的相对位置关系，将预设距离范围内的点云进行聚合处理时，具体用于：
    根据所述时序上相邻的点云之间的相对位置关系，将所述预设距离范围内的点云转换到同一个坐标系中。
  22. 根据权利要求13所述的***，其特征在于，所述处理器根据所述时序上相邻的点云之间的相对位置关系，对所述探测设备探测获得的点云进行聚合处理时，具体用于：
    根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理。
  23. 根据权利要求22所述的***，其特征在于，所述处理器根据所述时序上相邻的点云之间的相对位置关系，将预设时间范围内的点云进行聚合处理时，具体用于：
    根据所述时序上相邻的点云之间的相对位置关系，将所述预设时间范围内的点云转换到同一个坐标系中。
  24. 根据权利要求13-23任一项所述的***，其特征在于，所述探测设备在不同时刻探测获得的点云之间的目标相对位置关系，包括：
    所述探测设备探测获得的时序上不相邻的点云之间的目标相对位置关系。
  25. 一种可移动平台，其特征在于，包括：
    机身；
    动力***，安装在所述机身，用于提供移动动力；
    以及权利要求13-24任一项所述的点云处理***。
  26. 一种计算机可读存储介质，其特征在于，其上存储有计算机程序，所述计算机程序被处理器执行以实现权利要求1-12任一项所述的方法。
PCT/CN2019/104366 2019-09-04 2019-09-04 点云处理方法、***、可移动平台及存储介质 WO2021042286A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201980034379.9A CN112166460A (zh) 2019-09-04 2019-09-04 点云处理方法、***、可移动平台及存储介质
PCT/CN2019/104366 WO2021042286A1 (zh) 2019-09-04 2019-09-04 点云处理方法、***、可移动平台及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/104366 WO2021042286A1 (zh) 2019-09-04 2019-09-04 点云处理方法、***、可移动平台及存储介质

Publications (1)

Publication Number Publication Date
WO2021042286A1 true WO2021042286A1 (zh) 2021-03-11

Family

ID=73859407

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/104366 WO2021042286A1 (zh) 2019-09-04 2019-09-04 点云处理方法、***、可移动平台及存储介质

Country Status (2)

Country Link
CN (1) CN112166460A (zh)
WO (1) WO2021042286A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170305546A1 (en) * 2014-04-29 2017-10-26 Baidu Online Network Technology (Beijing) Co., Ltd. Autonomous navigation method and system, and map modeling method and system
CN108152831A (zh) * 2017-12-06 2018-06-12 中国农业大学 一种激光雷达障碍物识别方法及***
CN108267141A (zh) * 2016-12-30 2018-07-10 乐视汽车（北京）有限公司 道路点云数据处理***
CN108921947A (zh) * 2018-07-23 2018-11-30 百度在线网络技术（北京）有限公司 生成电子地图的方法、装置、设备、存储介质以及采集实体
CN109461207A (zh) * 2018-11-05 2019-03-12 胡翰 一种点云数据建筑物单体化方法及装置

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104463894B (zh) * 2014-12-26 2020-03-24 山东理工大学 一种多视角三维激光点云全局优化整体配准方法
US11353589B2 (en) * 2017-11-17 2022-06-07 Nvidia Corporation Iterative closest point process based on lidar with integrated motion estimation for high definition maps
CN109242828B (zh) * 2018-08-13 2022-04-19 浙江大学 基于光栅投影多步相移法的3d打印制品三维缺陷检测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170305546A1 (en) * 2014-04-29 2017-10-26 Baidu Online Network Technology (Beijing) Co., Ltd. Autonomous navigation method and system, and map modeling method and system
CN108267141A (zh) * 2016-12-30 2018-07-10 乐视汽车（北京）有限公司 道路点云数据处理***
CN108152831A (zh) * 2017-12-06 2018-06-12 中国农业大学 一种激光雷达障碍物识别方法及***
CN108921947A (zh) * 2018-07-23 2018-11-30 百度在线网络技术（北京）有限公司 生成电子地图的方法、装置、设备、存储介质以及采集实体
CN109461207A (zh) * 2018-11-05 2019-03-12 胡翰 一种点云数据建筑物单体化方法及装置

Also Published As

Publication number Publication date
CN112166460A (zh) 2021-01-01

Similar Documents

Publication Publication Date Title
EP4080248A1 (en) Method and apparatus for vehicle positioning, controller, smart car and system
WO2021072710A1 (zh) 移动物体的点云融合方法、***及计算机存储介质
US20230400585A1 (en) Fusion positioning method and apparatus, device and computer-readable storage medium
CN109407073B (zh) 反射值地图构建方法和装置
WO2018128674A1 (en) Systems and methods for mapping based on multi-journey data
JP6354556B2 (ja) 位置推定装置、位置推定方法、位置推定プログラム
WO2020133217A1 (zh) 连续障碍物检测方法、设备、***及存储介质
US10875178B2 (en) Motion target direction angle obtaining method, apparatus and robot using the same
CN112643664B (zh) 定位误差消除方法、装置、机器人及存储介质
CN112098964B (zh) 路端雷达的标定方法、装置、设备及存储介质
CN108326845B (zh) 基于双目相机和激光雷达的机器人定位方法、装置及***
CN111650598A (zh) 一种车载激光扫描***外参标定方法和装置
CN111380510B (zh) 重定位方法及装置、机器人
WO2021217518A1 (zh) 雷达点云聚类方法和装置
RU2767949C2 (ru) Способ (варианты) и система для калибровки нескольких лидарных датчиков
CN110906924A (zh) 一种定位初始化方法和装置、定位方法和装置及移动装置
CN114549738A (zh) 无人车室内实时稠密点云重建方法、***、设备及介质
WO2019135869A1 (en) Lidar intensity calibration
CN110930444B (zh) 一种基于双边优化的点云匹配方法、介质、终端和装置
CN115493601A (zh) 基于路网匹配的车辆自主定位方法、装置、存储介质
CN114820749A (zh) 无人车井下定位方法、***、设备及介质
KR101627419B1 (ko) 이동 노드의 위치 추정 방법 및 그 장치
WO2021081958A1 (zh) 地形检测方法、可移动平台、控制设备、***及存储介质
WO2024120187A1 (zh) 一种用于信息拒止环境下的无人机动态目标估计方法
JP7194217B2 (ja) 障害物速度確定方法、装置、電子機器、記憶媒体およびコンピュータプログラム

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: 19944174

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19944174

Country of ref document: EP

Kind code of ref document: A1