CN110750052A - Driving model training method and device, electronic equipment and medium - Google Patents
Driving model training method and device, electronic equipment and medium Download PDFInfo
- Publication number
- CN110750052A CN110750052A CN201910947160.XA CN201910947160A CN110750052A CN 110750052 A CN110750052 A CN 110750052A CN 201910947160 A CN201910947160 A CN 201910947160A CN 110750052 A CN110750052 A CN 110750052A
- Authority
- CN
- China
- Prior art keywords
- driving
- simulated
- target
- model
- controllable
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 80
- 238000012549 training Methods 0.000 title claims abstract description 58
- 230000002159 abnormal effect Effects 0.000 claims abstract description 21
- 230000003068 static effect Effects 0.000 claims abstract description 14
- 238000011156 evaluation Methods 0.000 claims abstract description 11
- 238000004088 simulation Methods 0.000 claims description 102
- 230000006399 behavior Effects 0.000 claims description 38
- 238000005286 illumination Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 abstract description 13
- 238000010586 diagram Methods 0.000 description 8
- 230000007613 environmental effect Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011022 operating instruction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B17/00—Systems involving the use of models or simulators of said systems
- G05B17/02—Systems involving the use of models or simulators of said systems electric
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Toys (AREA)
Abstract
The embodiment of the invention discloses a training method and a device of a driving model, electronic equipment and a medium, wherein the method comprises the following steps: generating a simulated driving environment of a simulated target, generating a simulated obstacle in the simulated driving environment, wherein the simulated obstacle comprises a dynamic obstacle and/or a static obstacle, acquiring running state information and driving environment information of the simulated target through a sensor on the simulated target, inputting the running state information and the driving environment information into a driving model, outputting driving control information by the driving model to control the driving of the simulated target, evaluating the driving behavior of the simulated target according to whether the driving of the simulated target is abnormal or not, and training the driving model according to an evaluation result. The embodiment of the invention can simulate a more real driving environment and simulate the actual driving behavior of the target, so that when the trained driving model is used for the automatic driving technology, the driving behavior of the target can be more accurately controlled, thereby improving the safety of automatic driving.
Description
Technical Field
The invention relates to an artificial intelligence technology, in particular to a training method and device of a driving model, electronic equipment and a medium.
Background
With the development of scientific technology, the unmanned technology is beginning to slowly move into people's lives. The unmanned technology is that the vehicle-mounted sensor is used for sensing the surrounding environment of the vehicle, and the steering and the speed of the vehicle are controlled according to the road, the vehicle position and the obstacle information obtained by sensing, so that the vehicle can safely and reliably run on the road.
However, existing unmanned technologies, such as multi-sensor based unmanned vehicle technologies, have many potential problems, and the cost required to slowly find out these problems through drive tests during actual driving is quite high. At this time, a simulation platform is needed to simulate the whole process of the unmanned driving, so as to find out the defects of the existing unmanned driving technology and improve the existing unmanned driving technology with less cost.
Disclosure of Invention
The embodiment of the invention provides a training technology of a driving model.
According to an aspect of an embodiment of the present invention, there is provided a training method of a driving model, including:
generating a simulated driving environment of a simulated target;
generating simulated obstacles in the simulated driving environment, the simulated obstacles comprising dynamic obstacles and/or static obstacles;
acquiring running state information and driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into a driving model, and outputting driving control information by the driving model to control the driving of the simulation target;
and evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not, and training the driving model according to an evaluation result.
Optionally, in each of the above method embodiments of the present invention, the simulation target includes: vehicles, robots, toy vehicles.
Optionally, in each of the above method embodiments of the present invention, the generating a simulated driving environment of a simulated target includes:
and constructing urban traffic and road traffic environments of the positions of the simulation targets, simulating weather and time, and simulating the positions and behaviors of random targets.
Optionally, in the foregoing method embodiments of the present invention, the constructing an urban traffic and road traffic environment of the position where the simulation target is located includes: constructing urban traffic and road traffic environments of the positions of the simulation targets through urban and road models; and/or the presence of a gas in the gas,
the simulated weather and time includes: simulating the weather condition of the city where the simulation target is located through a weather model, and simulating the current time and the illumination condition through a time model; and/or the presence of a gas in the gas,
the position and the behavior of the simulated random target comprise: and simulating the position, behavior and movement route of the random target according to the preset configuration information.
Optionally, in the above method embodiments of the present invention, the generating a simulated obstacle in the simulated driving environment includes:
receiving an obstacle generation request, the obstacle generation request comprising: obstacle type and number, generation time and geographical location;
and calling an obstacle model to generate a simulated obstacle in the simulated driving environment according to the obstacle generation request.
Optionally, in each of the above method embodiments of the present invention, after the generating the simulated driving environment of the simulated target, the method further includes:
receiving a controllable target operation instruction, wherein the controllable target operation instruction comprises: the type and number, location and travel route of the controllable objects;
and simulating a controllable target and an operation behavior in the simulated driving environment according to the controllable target operation instruction.
Optionally, in each of the above method embodiments of the present invention, the controllable target includes any one or more of the following: controllable pedestrians, controllable vehicles and controllable robots.
Optionally, in each of the above method embodiments of the present invention, the controllable target execution instruction further includes: the appearance and/or size of the object can be controlled.
Optionally, in each of the above method embodiments of the present invention, after simulating a controllable target and an operation behavior in the simulated driving environment according to the controllable target operation instruction, the method further includes:
receiving environmental information collected by a sensor on the controllable target;
and generating operation control information of the controllable target according to the environment information acquired by the sensor on the controllable target so as to control the operation of the controllable target based on the operation control information.
Optionally, in each of the above method embodiments of the present invention, the method further includes:
collecting traffic information of the position of the simulation target through a sensor on the simulation target and inputting the traffic information into the driving model;
the driving model determines a driving route of the simulation target based on the traffic information and the map of the simulated driving environment;
the outputting, by the driving model, driving control information includes: the driving model outputs driving control information based on the running state information, the driving environment information, and the travel route.
Optionally, in each of the above method embodiments of the present invention, whether the driving of the simulation target is abnormal includes any one or more of the following:
whether the simulation target deviates from a lane, whether the simulation target collides with an obstacle or not and whether the simulation target deviates from a driving route or not during driving.
According to another aspect of the embodiments of the present invention, there is provided a training apparatus for a driving model, including:
the scene generation module is used for generating a simulated driving environment of a simulated target; and invoking an obstacle model to generate simulated obstacles in the simulated driving environment, the simulated obstacles comprising dynamic obstacles and/or static obstacles;
the control module is used for acquiring the running state information and the driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into a driving model, and outputting driving control information by the driving model to control the driving of the simulation target;
and the training module is used for evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not and training the driving model according to an evaluation result.
Optionally, in each of the above method embodiments of the present invention, the simulation target includes: vehicles, robots, toy vehicles.
Optionally, in each of the method embodiments of the present invention, the scene generation module is specifically configured to: and constructing urban traffic and road traffic environments of the positions of the simulation targets, simulating weather and time, and simulating the positions and behaviors of random targets.
Optionally, in each of the method embodiments of the present invention, the scene generation module is specifically configured to: calling city and road models, and constructing city traffic and road traffic environments of the positions of the simulation targets; calling a weather model, simulating the weather condition of the city where the simulation target is located, and simulating the current time and the illumination condition through a time model; simulating the position, behavior and movement route of a random target according to preset configuration information;
the device further comprises:
the city and road model is used for constructing city traffic and road traffic environments of the positions of the simulation targets;
and the weather model is used for simulating the weather condition of the city where the simulation target is located, and the current time and the illumination condition are simulated through the time model.
Optionally, in each of the above method embodiments of the present invention, the obstacle model is specifically configured to: receiving an obstacle generation request, the obstacle generation request comprising: obstacle type and number, generation time and geographical location; and calling an obstacle model to generate a simulated obstacle in the simulated driving environment according to the obstacle generation request.
Optionally, in each of the above method embodiments of the present invention, the control module is further configured to receive a controllable target operation instruction, where the controllable target operation instruction includes: and simulating the controllable targets and the operation behaviors in the simulated driving environment according to the controllable target operation instructions.
Optionally, in each of the above method embodiments of the present invention, the controllable target includes any one or more of the following: controllable pedestrians, controllable vehicles and controllable robots.
Optionally, in each of the above method embodiments of the present invention, the controllable target execution instruction further includes: the appearance and/or size of the object can be controlled.
Optionally, in each of the above method embodiments of the present invention, the method further includes:
the controllable target generation module is used for receiving environmental information acquired by a sensor on the controllable target; and generating operation control information of the controllable target according to the environment information acquired by the sensor on the controllable target so as to control the operation of the controllable target based on the operation control information.
Optionally, in each of the above method embodiments of the present invention, the control module is further configured to acquire traffic information of a location where the simulation target is located through a sensor on the simulation target and input the driving model, so that the driving model determines the driving route of the simulation target based on the traffic information and the map of the simulated driving environment.
Optionally, in each of the above method embodiments of the present invention, the method further includes:
a driving model for outputting driving control information based on the operation state information, the driving environment information, and the driving route.
Optionally, in each of the above method embodiments of the present invention, whether the driving of the simulation target is abnormal includes any one or more of the following: whether the simulation target deviates from a lane, whether the simulation target collides with an obstacle or not and whether the simulation target deviates from a driving route or not during driving.
According to still another aspect of the embodiments of the present invention, there is provided an electronic apparatus including: a processor and a memory;
the memory is used for storing at least one executable instruction, and the executable instruction enables the processor to execute the operation of each step in the training method of the driving model according to any one of the above embodiments of the invention.
According to another aspect of the embodiments of the present invention, there is provided a computer-readable medium for storing computer-readable instructions, which when executed, implement the operations of the steps in the training method of the driving model according to any one of the above embodiments of the present invention.
Based on the training method and device of the driving model, the electronic device and the medium provided by the embodiment of the invention, after the simulated driving environment of the simulated target is generated, the simulated obstacle including the dynamic obstacle and/or the static obstacle can be generated in the simulated driving environment, the running state information and the driving environment information of the simulated target are acquired through the sensor on the simulated target and are input into the driving model, the driving control information is output by the driving model to control the driving of the simulated target, the driving behavior of the simulated target is evaluated according to whether the driving of the simulated target is abnormal or not, and the driving model is trained according to the evaluation result. The embodiment of the invention can simulate a more real driving environment and the actual driving behavior of the simulation target, and train the driving model according to whether the driving of the simulation target is abnormal, so that when the trained driving model is used for an automatic driving technology, the driving behavior of the target can be more accurately controlled, thereby improving the safety of automatic driving.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description, serve to explain the principles of the invention.
The invention will be more clearly understood from the following detailed description, taken with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of a method for training a driving model according to an embodiment of the present invention.
FIG. 2 is a flowchart of another embodiment of a method for training a driving model of the present invention.
FIG. 3 is a flow chart of a method for training a driving model according to another embodiment of the present invention.
FIG. 4 is a flowchart of a method for training a driving model according to yet another embodiment of the present invention.
Fig. 5 is a schematic structural diagram of an embodiment of the training device for the driving model of the present invention.
Fig. 6 is a schematic structural diagram of another embodiment of the training device for the driving model of the present invention.
Fig. 7 is an exemplary block diagram of an electronic device embodiment of the present invention.
Detailed Description
Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that: the relative arrangement of the components and steps, the numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.
Meanwhile, it should be understood that the sizes of the respective portions shown in the drawings are not drawn in an actual proportional relationship for the convenience of description.
The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.
Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail but are intended to be part of the specification where appropriate.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, further discussion thereof is not required in subsequent figures.
Embodiments of the invention are operational with numerous other general purpose or special purpose computing system environments or configurations, and with numerous other electronic devices, such as terminal devices, computer systems, servers, etc. Examples of well known terminal devices, computing systems, environments, and/or configurations that may be suitable for use with electronic devices, such as terminal devices, computer systems, servers, and the like, include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, hand-held or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, networked personal computers, minicomputer systems, mainframe computer systems, distributed cloud computing environments that include any of the above, and the like.
Electronic devices such as terminal devices, computer systems, servers, etc. may be described in the general context of computer system-executable instructions, such as program modules, being executed by a computer system. Generally, program modules may include routines, programs, objects, components, logic, data structures, etc. that perform particular tasks or implement particular abstract data types. The computer system/server may be practiced in distributed cloud computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computer system storage media including memory storage devices.
FIG. 1 is a flow chart of a method for training a driving model according to an embodiment of the present invention. As shown in fig. 1, the training method of the driving model of the embodiment includes:
102, generating a simulated driving environment of the simulated target.
In some implementations of embodiments of the invention, the simulation target may be any object that may be exercised or moved, and may include, for example and without limitation, any one or more of the following: vehicles, robots, toy vehicles, and the like. The present invention is not limited to the specific representation of the simulation target.
Simulated obstacles are generated in the simulated driving environment, the simulated obstacles including dynamic obstacles and/or static obstacles 104.
And 106, acquiring the running state information and the driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into the driving model, and outputting driving control information by the driving model to control the running of the simulation target.
The operation state information of the simulation target may include, but is not limited to: speed of movement, acceleration of movement, direction of travel, and the like. The driving environment information of the simulated target may include, for example, but is not limited to: and simulating urban roads, obstacles and other information in the driving environment. The driving control information is control information for controlling the operation of the simulation target, and may include, for example and without limitation: information of a moving speed, a moving acceleration, whether to change lanes, whether to turn on/off lights, and the like of the simulation target is controlled.
And 108, evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not, and training the driving model according to an evaluation result, so that the driving model obtained after the final training is not abnormal in the process of controlling the driving of the simulation target.
In some implementations of the embodiments of the present invention, whether the driving of the simulation target is abnormal may include, but is not limited to, any one or more of the following: whether the simulation target deviates from a lane, whether the simulation target collides with an obstacle, whether the simulation target deviates from a driving route and the like during driving. If any one or more behaviors occur in the simulation target, the driving of the simulation target is considered to be abnormal. The present invention does not limit the specific type of abnormality occurring in the running of the simulation target.
Based on the training method of the driving model provided by the above embodiment of the present invention, after the simulated driving environment of the simulated target is generated, the simulated obstacle including the dynamic obstacle and/or the static obstacle can be generated in the simulated driving environment, the running state information and the driving environment information of the simulated target are collected by the sensor on the simulated target, and are input into the driving model, the driving model outputs the driving control information to control the driving of the simulated target, the driving behavior of the simulated target is evaluated according to whether the driving of the simulated target is abnormal, and the driving model is trained according to the evaluation result. The embodiment of the invention can simulate a more real driving environment and the actual driving behavior of the simulation target, and train the driving model according to whether the driving of the simulation target is abnormal, so that when the trained driving model is used for an automatic driving technology, the driving behavior of the target can be more accurately controlled, thereby improving the safety of automatic driving.
In an alternative example of the embodiments of the present invention, in operation 102, generating a simulated driving environment of the simulated target may include, for example: the method comprises the steps of constructing urban traffic and road traffic environments of positions of simulation targets, simulating weather and time, and simulating positions and behaviors of random targets.
In some optional examples, city traffic and road traffic environments where the simulation target is located may be constructed through city and road models to simulate city traffic. The city model may include houses, billboards, roadside decorations, and the like, among others. The road model may include urban roads and expressways, traffic signals and signs, and so on. For example, the road model may simulate each road curvature, the number of lanes involved, lane lines, the width of each lane, the number of other vehicles on each lane, the direction and speed of travel, lane speed limits, the number of pedestrians crossing the road, the speed, the direction of traffic flow simulating the lane in which the vehicle is located, lane speed limits, the status of intersection traffic lights, the duration of each traffic light signal.
In some alternative examples, the weather conditions of the city in which the simulation target is located may be simulated by a weather model (e.g., sunny, cloudy, rainy, windy, snowy), and the current time and lighting conditions may be simulated by a time model (e.g., morning of the day, afternoon of the day, night).
In some alternative examples, when simulating the position and behavior of the random object, the position, behavior and movement route of the random object may be simulated according to preset configuration information. The random target is an object that can randomly appear, move and disappear in the simulated driving environment, and may include, but is not limited to: random pedestrians, random vehicles, random robots, and the like. When the random target is a random pedestrian, the appearance (such as clothes color and age) of the random pedestrian can be simulated according to preset configuration information; when the random object is a random vehicle, the appearance of the random vehicle (e.g., the shape, color, etc. of the vehicle) may also be simulated in accordance with preset configuration information.
Based on the embodiment, the simulated obstacles including dynamic obstacles and/or static obstacles can be generated in the simulated driving environment, the urban traffic and road traffic environment of the position of the simulated target can be constructed, the weather and time can be simulated, and the position and the behavior of the random target can be simulated, so that a more practical simulated driving environment can be provided, the training of the driving model can be realized in various possible simulated driving environments, and the driving model obtained through training can correctly control the running of the target object in various driving environments.
FIG. 2 is a flowchart of another embodiment of a method for training a driving model of the present invention. As shown in fig. 2, based on the embodiment shown in fig. 1, operation 104 includes:
202, receiving an obstacle generation request, the obstacle generation request comprising: obstacle type and number, generation time and geographical location.
Among them, the obstacle types may include static obstacles such as scattered garbage, leaves, plastic bags, stones, etc., and dynamic obstacles such as fallen garbage, flying birds, running animals, rolling balls, etc.
And 204, calling an obstacle model to generate a simulated obstacle in the simulated driving environment according to the obstacle generation request.
The method comprises the steps of calling an external obstacle model through a Remote Procedure Call (RPC) communication mode, and sending an obstacle generation request to the obstacle model to generate a simulated obstacle in the simulated driving environment.
Based on the embodiment, the obstacle model can be called to generate the simulated obstacles in the simulated driving environment, including static obstacles and dynamic obstacles, so that the simulated driving environment is more suitable for the actual situation, the difficulty of simulated driving is increased, and the driving model obtained by training has better driving performance.
FIG. 3 is a flow chart of a method for training a driving model according to another embodiment of the present invention. As shown in fig. 3, on the basis of the embodiment shown in fig. 1 or fig. 2, after operation 102, the method may further include:
receiving a controllable target operation instruction 302, where the controllable target operation instruction may include: the type and number of controllable objects, location and travel route.
In some implementations of embodiments of the present invention, the controllable target is an object in the simulated driving environment that can control its operation, and may include, for example and without limitation, any one or more of the following: controllable pedestrians, controllable vehicles, controllable robots, and the like. The invention is not limited to the specific representation of the controllable object.
In addition, optionally, in other optional examples, the controllable target execution instruction may further include: the appearance and/or size of the object can be controlled. For example, when the controllable object is a controllable pedestrian, the controllable object operating instructions may also include the appearance and/or size of the controllable pedestrian (e.g., clothing color, age, disability, etc.); when the controllable object is a controllable vehicle, the controllable object operating instructions may also include the appearance of the controllable vehicle (e.g., the shape, color, etc. of the vehicle).
And 304, simulating the controllable target and the running behavior thereof in the simulated driving environment according to the controllable target running instruction.
Further, optionally, referring back to fig. 3, after the operation 304, the method may further include:
and 306, receiving environmental information collected by a sensor on the controllable target.
In the embodiment of the invention, one or more sensors can be arranged on each controllable target and used for acquiring the environmental information around the controllable target in real time; the collected environment information can be transmitted to an external program through an RPC communication mode so as to assist the external program to complete the operation of a subsequent controllable target. The external program may be a driving model or a separately formed controllable target control model.
308, generating operation control information for the controllable target according to the environmental information collected by the sensor on the controllable target, so as to control the operation of the controllable target based on the operation control information.
Based on the embodiment, various possible controllable targets (such as vehicles, pedestrians and the like) can be simulated in the simulated driving environment, the operation of the controllable targets can be controlled according to the environmental information collected by the sensors on the controllable targets, the simulated driving environment is more suitable for the actual situation, the difficulty of simulated driving is increased, and the trained driving model has better driving performance.
FIG. 4 is a flowchart of a method for training a driving model according to yet another embodiment of the present invention. As shown in fig. 4, this embodiment includes:
a simulated driving environment is generated that simulates the target 402.
In an alternative example of embodiments of the present invention, in operation 402, city traffic and road traffic environments may be constructed that simulate the location of the target, simulate weather and time, and simulate the location and behavior of random targets.
A simulated obstacle is generated 404 in the simulated driving environment, the simulated obstacle including a dynamic obstacle and/or a static obstacle.
And 406, acquiring the running state information and the driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into the driving model, and acquiring the traffic information (such as the traffic light condition, the positions and the running states of other targets on the road, the destination and the like) of the position of the simulation target through a sensor on the simulation target, and inputting the traffic information into the driving model.
The traffic information of the position of the simulation target may include, but is not limited to: the next traffic light status of the lane where the simulation target is located, lane information at the curve, vehicle information at the curve, traffic accident, and the like.
The driving model determines 408 a driving route of the simulation target based on the traffic information and the map of the simulated driving environment.
The map for simulating the driving environment can be generated in advance and stored in the driving model or a storage module which can be called by the driving model.
The driving model outputs driving control information based on the above-described operating state information, driving environment information, and driving route to control the driving of the simulation target 410.
And 412, evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not, and training a driving model according to an evaluation result.
Based on the embodiment, a simulated driving environment of the simulated target can be generated, the running state information and the driving environment information of the simulated target are collected through the sensor on the simulated target and input into the driving model, the traffic information of the position of the simulated target is collected through the sensor on the simulated target and input into the driving model, the driving route of the simulated target is determined by the driving model, the driving control information is output to control the driving of the simulated target based on the running state information, the driving environment information and the driving route, and then the driving model is trained according to whether the driving of the simulated target is abnormal, so that the trained driving model can select a correct and reasonable driving route to control the target (such as a vehicle, a robot and the like) to safely drive.
Fig. 5 is a schematic structural diagram of an embodiment of the training device for the driving model of the present invention. The training device of this embodiment can be used to implement the above-described embodiments of the training method of the present invention. As shown in fig. 5, the training device of the driving model of the embodiment includes: the system comprises a scene generation module, a control module and a training module. Wherein:
a scenario generation module, configured to generate a simulated driving environment of a simulated target, where the simulated target may include, for example: vehicles, robots, toy vehicles, and the like; and invoking the obstacle model to generate a simulated obstacle in the simulated driving environment, the simulated obstacle comprising a dynamic obstacle and/or a static obstacle.
The control module is used for acquiring the running state information and the driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into a driving model, and outputting driving control information by the driving model to control the driving of the simulation target;
and the training module is used for evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not and training the driving model according to an evaluation result.
In some embodiments, whether the driving of the simulation target is abnormal may include, but is not limited to, any one or more of the following: whether the simulation target deviates from a lane, whether the simulation target collides with an obstacle, whether the simulation target deviates from a driving route and the like during driving.
Based on the training device of the driving model provided by the above embodiment of the invention, after the simulated driving environment of the simulated target is generated, the simulated obstacle including the dynamic obstacle and/or the static obstacle can be generated in the simulated driving environment, the running state information and the driving environment information of the simulated target are acquired by the sensor on the simulated target and are input into the driving model, the driving model outputs the driving control information to control the driving of the simulated target, the driving behavior of the simulated target is evaluated according to whether the driving of the simulated target is abnormal or not, and the driving model is trained according to the evaluation result. The embodiment of the invention can simulate a more real driving environment and the actual driving behavior of the simulation target, and train the driving model according to whether the driving of the simulation target is abnormal, so that when the trained driving model is used for an automatic driving technology, the driving behavior of the target can be more accurately controlled, thereby improving the safety of automatic driving.
In some embodiments, the scene generation module is specifically configured to construct an urban traffic and road traffic environment that simulates the location of the target, simulate weather and time, and simulate the location and behavior of the random target.
For example, in some optional examples, the scene generation module is specifically configured to: calling city and road models, and constructing city traffic and road traffic environments of the positions of the simulation targets; calling a weather model, simulating the weather condition of the city where the simulation target is located, and simulating the current time and the illumination condition through a time model; and simulating the position, behavior and movement route of the random target according to the preset configuration information.
Fig. 6 is a schematic structural diagram of another embodiment of the training device for the driving model of the present invention. As shown in fig. 6, on the basis of fig. 5, the training device for the driving model may further include: the urban and road model is used for constructing urban traffic and road traffic environments of the positions of the simulation targets; and the weather model is used for simulating the weather condition of the city where the simulation target is located, and the current time and the illumination condition are simulated through the time model.
In some of these embodiments, the obstacle model is specifically used to: receiving an obstacle generation request, the obstacle generation request including: obstacle type and number, generation time and geographical location; and calling the obstacle model to generate a simulated obstacle in the simulated driving environment according to the obstacle generation request.
In another embodiment of the training apparatus for driving model of the present invention, the control module is further configured to receive a controllable target operation command, where the controllable target operation command includes: the type and number of controllable objects, location and travel route, and optionally the appearance and/or size of the controllable objects; and simulating the controllable target and the operation behavior in the driving environment according to the controllable target operation instruction.
In some embodiments, the controllable target may include, but is not limited to, any one or more of the following: controllable pedestrians, controllable vehicles and controllable robots.
In addition, the training apparatus according to the embodiment of the present invention may further include: the controllable target generation module is used for receiving environmental information acquired by a sensor on a controllable target; and generating operation control information for the controllable target according to the environmental information acquired by the sensor on the controllable target so as to control the operation of the controllable target based on the operation control information.
In another embodiment of the training apparatus for a driving model according to the present invention, the control module may be further configured to collect traffic information of a location of the simulation target through a sensor on the simulation target and input the traffic information into the driving model, so that the driving model determines a driving route of the simulation target based on the traffic information and a map of the simulated driving environment.
In addition, referring to fig. 6 again, the training apparatus according to the embodiment of the present invention may further include: and a driving model for outputting driving control information based on the running state information, the driving environment information, and the driving route.
In addition, an embodiment of the present invention further provides an electronic device, including: a processor and a memory;
the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the training method of the driving model according to any one of the above embodiments of the invention.
Fig. 7 is an exemplary block diagram of an electronic device embodiment of the present invention. Next, an electronic apparatus according to an embodiment of the present disclosure is described with reference to fig. 7. The electronic device may be either or both of the first device and the second device, or a stand-alone device separate from them, which stand-alone device may communicate with the first device and the second device to receive the acquired input signals therefrom. As shown in fig. 7, the electronic device includes one or more processors and memory.
The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. The processor may be configured to perform the process steps of the training method of any of the driving models of fig. 1-4.
The memory may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium and executed by a processor to implement the above-described methods of training of driving models of the various embodiments of the present disclosure and/or other desired functions.
In one example, the electronic device may further include: an input device and an output device, which are interconnected by a bus system and/or other form of connection mechanism (not shown).
The input device may also include, for example, a keyboard, a mouse, and the like.
The output device may output various information including the determined distance information, direction information, and the like to the outside. The output devices may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, among others.
Of course, for simplicity, only some of the components of the electronic device relevant to the present disclosure are shown in fig. 7, omitting components such as buses, input/output interfaces, and the like. In addition, the electronic device may include any other suitable components, depending on the particular application.
In addition to the above-described methods and apparatus, embodiments of the present disclosure may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the training method of the driving model of the above-described embodiments of the present disclosure.
The computer program product may write program code for carrying out operations for embodiments of the present disclosure in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present disclosure may also include a computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform the steps in the training method of the driving model of the above-described embodiments of the present disclosure.
The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing describes the general principles of the present disclosure in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present disclosure are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present disclosure. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the disclosure is not intended to be limited to the specific details so described.
In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts in the embodiments are referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The block diagrams of devices, apparatuses, systems referred to in this disclosure are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
The methods and apparatus of the present disclosure may be implemented in a number of ways. For example, the methods and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order for the steps of the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above unless specifically stated otherwise. Further, in some embodiments, the present disclosure may also be embodied as programs recorded in a recording medium, the programs including machine-readable instructions for implementing the methods according to the present disclosure. Thus, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.
It is also noted that in the devices, apparatuses, and methods of the present disclosure, each component or step can be decomposed and/or recombined. These decompositions and/or recombinations are to be considered equivalents of the present disclosure.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit embodiments of the disclosure to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. A method of training a driving model, comprising:
generating a simulated driving environment of a simulated target;
generating simulated obstacles in the simulated driving environment, the simulated obstacles comprising dynamic obstacles and/or static obstacles;
acquiring running state information and driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into a driving model, and outputting driving control information by the driving model to control the driving of the simulation target;
and evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not, and training the driving model according to an evaluation result.
2. The method of claim 1, wherein simulating the target comprises: vehicles, robots, toy vehicles.
3. The method of claim 1 or 2, wherein the generating a simulated driving environment of a simulated target comprises:
and constructing urban traffic and road traffic environments of the positions of the simulation targets, simulating weather and time, and simulating the positions and behaviors of random targets.
4. The method of claim 3, wherein constructing the urban and road traffic environment of the location of the simulated target comprises: constructing urban traffic and road traffic environments of the positions of the simulation targets through urban and road models; and/or the presence of a gas in the gas,
the simulated weather and time includes: simulating the weather condition of the city where the simulation target is located through a weather model, and simulating the current time and the illumination condition through a time model; and/or the presence of a gas in the gas,
the position and the behavior of the simulated random target comprise: and simulating the position, behavior and movement route of the random target according to the preset configuration information.
5. The method of any of claims 1-4, wherein generating a simulated obstacle in the simulated driving environment comprises:
receiving an obstacle generation request, the obstacle generation request comprising: obstacle type and number, generation time and geographical location;
and calling an obstacle model to generate a simulated obstacle in the simulated driving environment according to the obstacle generation request.
6. The method of any of claims 1-5, wherein after generating the simulated driving environment of the simulated target, further comprising:
receiving a controllable target operation instruction, wherein the controllable target operation instruction comprises: the type and number, location and travel route of the controllable objects;
and simulating a controllable target and an operation behavior in the simulated driving environment according to the controllable target operation instruction.
7. The method of claim 6, wherein the controllable objective comprises any one or more of: controllable pedestrians, controllable vehicles and controllable robots.
8. A training apparatus for a driving model, comprising:
the scene generation module is used for generating a simulated driving environment of a simulated target; and invoking an obstacle model to generate simulated obstacles in the simulated driving environment, the simulated obstacles comprising dynamic obstacles and/or static obstacles;
the control module is used for acquiring the running state information and the driving environment information of the simulation target through a sensor on the simulation target, inputting the running state information and the driving environment information into a driving model, and outputting driving control information by the driving model to control the driving of the simulation target;
and the training module is used for evaluating the driving behavior of the simulation target according to whether the driving of the simulation target is abnormal or not and training the driving model according to an evaluation result.
9. An electronic device, comprising: a processor and a memory;
the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation of each step in the training method of the driving model according to any one of claims 1-7.
10. A computer-readable medium storing computer-readable instructions, wherein the instructions, when executed, implement the operations of the steps of the training method of the driving model according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910947160.XA CN110750052A (en) | 2019-09-30 | 2019-09-30 | Driving model training method and device, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910947160.XA CN110750052A (en) | 2019-09-30 | 2019-09-30 | Driving model training method and device, electronic equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110750052A true CN110750052A (en) | 2020-02-04 |
Family
ID=69277685
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910947160.XA Pending CN110750052A (en) | 2019-09-30 | 2019-09-30 | Driving model training method and device, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110750052A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111538335A (en) * | 2020-05-15 | 2020-08-14 | 深圳国信泰富科技有限公司 | Anti-collision method of driving robot |
| CN111708644A (en) * | 2020-05-29 | 2020-09-25 | 北京百度网讯科技有限公司 | Virtual world management method and system for automatic driving simulation |
| CN111899625A (en) * | 2020-07-16 | 2020-11-06 | 北京理工大学 | Intelligent driving assisting development device |
| CN112373482A (en) * | 2020-11-23 | 2021-02-19 | 浙江天行健智能科技有限公司 | Driving habit modeling method based on driving simulator |
| CN112590813A (en) * | 2020-12-09 | 2021-04-02 | 禾多科技（北京）有限公司 | Method, apparatus, electronic device, and medium for generating information of autonomous vehicle |
| WO2022156520A1 (en) * | 2021-01-25 | 2022-07-28 | 国汽智控(北京)科技有限公司 | Cloud-road collaborative automatic driving model training method and system, and cloud-road collaborative automatic driving model calling method and system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107807542A (en) * | 2017-11-16 | 2018-03-16 | 北京北汽德奔汽车技术中心有限公司 | Automatic Pilot analogue system |
| CN108027243A (en) * | 2016-07-21 | 2018-05-11 | 百度（美国）有限责任公司 | For operating the control error correction planing method of automatic driving vehicle |
| CN108230817A (en) * | 2017-11-30 | 2018-06-29 | 商汤集团有限公司 | Vehicle drive analogy method and device, electronic equipment, system, program and medium |
| CN108931927A (en) * | 2018-07-24 | 2018-12-04 | 百度在线网络技术（北京）有限公司 | The creation method and device of unmanned simulating scenes |
| CN108983633A (en) * | 2017-06-02 | 2018-12-11 | 北京图森未来科技有限公司 | Simulation system, the method and apparatus of more objects in a kind of virtual environment |
| CN108982117A (en) * | 2018-07-04 | 2018-12-11 | 长安大学 | A kind of unmanned vehicle test macro and method |
| CN109643125A (en) * | 2016-06-28 | 2019-04-16 | 柯尼亚塔有限公司 | For training the 3D virtual world true to nature of automated driving system to create and simulation |
| CN109782737A (en) * | 2019-01-28 | 2019-05-21 | 平安科技（深圳）有限公司 | Unmanned vehicle analog detection method and system |
| CN109884916A (en) * | 2019-02-26 | 2019-06-14 | 初速度（苏州）科技有限公司 | A kind of automatic Pilot Simulation Evaluation method and device |
| US20190278280A1 (en) * | 2016-12-15 | 2019-09-12 | Hitachi Automotive Systems, Ltd. | Vehicle control device |
-
2019
- 2019-09-30 CN CN201910947160.XA patent/CN110750052A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109643125A (en) * | 2016-06-28 | 2019-04-16 | 柯尼亚塔有限公司 | For training the 3D virtual world true to nature of automated driving system to create and simulation |
| CN108027243A (en) * | 2016-07-21 | 2018-05-11 | 百度（美国）有限责任公司 | For operating the control error correction planing method of automatic driving vehicle |
| US20190278280A1 (en) * | 2016-12-15 | 2019-09-12 | Hitachi Automotive Systems, Ltd. | Vehicle control device |
| CN108983633A (en) * | 2017-06-02 | 2018-12-11 | 北京图森未来科技有限公司 | Simulation system, the method and apparatus of more objects in a kind of virtual environment |
| CN107807542A (en) * | 2017-11-16 | 2018-03-16 | 北京北汽德奔汽车技术中心有限公司 | Automatic Pilot analogue system |
| CN108230817A (en) * | 2017-11-30 | 2018-06-29 | 商汤集团有限公司 | Vehicle drive analogy method and device, electronic equipment, system, program and medium |
| CN108982117A (en) * | 2018-07-04 | 2018-12-11 | 长安大学 | A kind of unmanned vehicle test macro and method |
| CN108931927A (en) * | 2018-07-24 | 2018-12-04 | 百度在线网络技术（北京）有限公司 | The creation method and device of unmanned simulating scenes |
| CN109782737A (en) * | 2019-01-28 | 2019-05-21 | 平安科技（深圳）有限公司 | Unmanned vehicle analog detection method and system |
| CN109884916A (en) * | 2019-02-26 | 2019-06-14 | 初速度（苏州）科技有限公司 | A kind of automatic Pilot Simulation Evaluation method and device |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111538335A (en) * | 2020-05-15 | 2020-08-14 | 深圳国信泰富科技有限公司 | Anti-collision method of driving robot |
| CN111708644A (en) * | 2020-05-29 | 2020-09-25 | 北京百度网讯科技有限公司 | Virtual world management method and system for automatic driving simulation |
| CN111899625A (en) * | 2020-07-16 | 2020-11-06 | 北京理工大学 | Intelligent driving assisting development device |
| CN112373482A (en) * | 2020-11-23 | 2021-02-19 | 浙江天行健智能科技有限公司 | Driving habit modeling method based on driving simulator |
| CN112373482B (en) * | 2020-11-23 | 2021-11-05 | 浙江天行健智能科技有限公司 | Driving habit modeling method based on driving simulator |
| CN112590813A (en) * | 2020-12-09 | 2021-04-02 | 禾多科技（北京）有限公司 | Method, apparatus, electronic device, and medium for generating information of autonomous vehicle |
| CN112590813B (en) * | 2020-12-09 | 2022-09-27 | 禾多科技（北京）有限公司 | Method, device, electronic device and medium for generating information of automatic driving vehicle |
| WO2022156520A1 (en) * | 2021-01-25 | 2022-07-28 | 国汽智控(北京)科技有限公司 | Cloud-road collaborative automatic driving model training method and system, and cloud-road collaborative automatic driving model calling method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108230817B (en) | Vehicle driving simulation method and apparatus, electronic device, system, program, and medium | |
| CN110750052A (en) | Driving model training method and device, electronic equipment and medium | |
| Chen et al. | A novel integrated simulation and testing platform for self-driving cars with hardware in the loop | |
| CN112703459B (en) | Iterative generation of confrontational scenarios | |
| CN115843347A (en) | Generating autonomous vehicle simulation data from recorded data | |
| US20180211120A1 (en) | Training An Automatic Traffic Light Detection Model Using Simulated Images | |
| JP2019215849A (en) | Method, device, apparatus, and medium for classifying driving scene data | |
| US11989116B1 (en) | Collision evaluation for log-based simulations | |
| US20230150529A1 (en) | Dynamic sensor data augmentation via deep learning loop | |
| US11740620B2 (en) | Operational testing of autonomous vehicles | |
| Haavaldsen et al. | Autonomous vehicle control: End-to-end learning in simulated urban environments | |
| CN109461342B (en) | Teaching system for unmanned motor vehicle and teaching method thereof | |
| CN112740188A (en) | Log-based simulation using biases | |
| CN113064487A (en) | Interactive virtual simulation system | |
| KR102589587B1 (en) | Dynamic model evaluation package for autonomous driving vehicles | |
| CN116194350A (en) | Generating multiple simulated edge condition driving scenarios | |
| JP2022547611A (en) | Simulation of various long-term future trajectories in road scenes | |
| KR102157587B1 (en) | Simulation method for autonomous vehicle linked game severs | |
| CN114444208A (en) | Method, device, equipment and medium for determining reliability of automatic driving system | |
| Nikitin et al. | Traffic Signs Recognition System Development | |
| CN116710732A (en) | Rare event simulation in autonomous vehicle motion planning | |
| CN113029155A (en) | Robot automatic navigation method and device, electronic equipment and storage medium | |
| Rossi et al. | Vehicle hardware-in-the-loop system for ADAS virtual testing | |
| Komavec et al. | Simulation Environment for Testing Autonomous Vehicles | |
| US11834070B2 (en) | Probabilistic simulation sampling from agent data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200204 |