WO2023226733A1 - Procédé et appareil d'acquisition de données de scène de véhicule, support de stockage et dispositif électronique - Google Patents

Procédé et appareil d'acquisition de données de scène de véhicule, support de stockage et dispositif électronique Download PDF

Info

Publication number
WO2023226733A1
WO2023226733A1 PCT/CN2023/092610 CN2023092610W WO2023226733A1 WO 2023226733 A1 WO2023226733 A1 WO 2023226733A1 CN 2023092610 W CN2023092610 W CN 2023092610W WO 2023226733 A1 WO2023226733 A1 WO 2023226733A1
Authority
WO
WIPO (PCT)
Prior art keywords
vehicle
data
scene
target
determined
Prior art date
Application number
PCT/CN2023/092610
Other languages
English (en)
Chinese (zh)
Other versions
WO2023226733A9 (fr
Inventor
陈志新
陈博
尚秉旭
刘洋
王洪峰
张勇
Original Assignee
中国第一汽车股份有限公司
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 中国第一汽车股份有限公司 filed Critical 中国第一汽车股份有限公司
Publication of WO2023226733A1 publication Critical patent/WO2023226733A1/fr
Publication of WO2023226733A9 publication Critical patent/WO2023226733A9/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/20—Instruments for performing navigational calculations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads

Definitions

  • the present disclosure relates to the field of communications, and specifically, to a vehicle scene data collection method, device, storage medium and electronic device.
  • the autonomous driving algorithm needs to deal with many and complex scenarios.
  • the algorithm has a long life cycle and requires long-term iterative optimization.
  • the test is based on fixed test scenarios, and it is found that the autonomous driving algorithm cannot When covering the current working conditions, testers manually record problems, optimize the software, and then test, which is inefficient. There will also be some edge scenarios where the test cannot be recognized, resulting in the inability to discover all problems with the software. That is, the collection of existing scene data is all artificially simulated scenes, and testers manually trigger the collection of scenes, resulting in less scene data that can be obtained and lack of authenticity.
  • the purpose of the present disclosure is to provide a vehicle scene data collection method, device, storage medium and electronic device, so as to at least solve the technical problem of less vehicle scene data being obtained.
  • the present disclosure provides a method for collecting vehicle scene data.
  • the method includes: during the driving process of the vehicle, acquiring target data determined by the automatic driving system of the vehicle, wherein:
  • the target data includes at least one of the following: control instructions of the vehicle by the automatic driving system, sensing data of the environment in which the vehicle is located by the automatic driving system, and the vehicle determined by the automatic driving system.
  • the positioning data and the prediction planning data predicted by the automatic driving system; determine whether the vehicle is in the target scene according to the target data; when it is determined that the vehicle is in the target scene, collect the location of the vehicle in the target scene.
  • Vehicle scene data in the target scene is acquiring target data determined by the automatic driving system of the vehicle, wherein:
  • the target data includes at least one of the following: control instructions of the vehicle by the automatic driving system, sensing data of the environment in which the vehicle is located by the automatic driving system, and the vehicle determined by the automatic driving system.
  • the positioning data and the prediction planning data predicted by the automatic driving system; determine whether the vehicle is in
  • determining whether the vehicle is in a target scene according to the target data includes: determining that the vehicle is in an autonomous driving mode, and When the control instruction indicates that the acceleration of the vehicle in the first direction is set to the first acceleration, it is determined that the vehicle is in an abnormal driving scene, wherein the first direction is the forward direction of the vehicle, so The value of the first acceleration exceeds the first threshold; after determining that the vehicle is in the automatic driving mode, the control instruction instructs to set the acceleration of the vehicle in the second direction to the second acceleration, and instructs to set the vehicle's tires to When the turning angle change rate is set as the target turning angle change rate, it is determined that the vehicle is in an abnormal driving scene, wherein the second acceleration exceeds a second threshold, the target turning angle changing rate exceeds a third threshold, and the second The angle between the direction and the second direction is a preset angle; when it is determined that the vehicle is in manual driving mode, determine the target control instruction issued by the target object to the vehicle; combine the target
  • determining whether the vehicle is in a target scene according to the target data includes: determining the automatic driving according to the perception data.
  • the type of the first obstacle sensed by the sensing module of the system changes within the first preset time period, it is determined that the vehicle is in an abnormal scene of the sensing module; when it is determined based on the sensing data that the identification of the first obstacle is in Occurs within the first preset time period If there is a change, it is determined that the vehicle is in an abnormal scene of the sensing module; if it is determined based on the sensing data that the change in the moving speed of the first obstacle within the first preset time period exceeds the fifth threshold, it is determined that the The vehicle is in an abnormal scene of the sensing module; when it is determined according to the sensing data that the position of the second obstacle sensed by the sensing module changes within a first preset time period, it is determined that the vehicle is in an abnormal scene of the sensing module; When it is determined based on the sensing data.
  • determining whether the vehicle is in a target scene according to the target data includes: determining the location of the vehicle according to the positioning data.
  • determining whether the vehicle is in an abnormal scene of the positioning module includes: determining the location of the vehicle according to the positioning data.
  • the change amount exceeds the seventh threshold it is determined that the vehicle is in an abnormal scene of the positioning module; it is determined based on the positioning data that the position of the vehicle has not changed within the second preset time period, but the When the speed of the vehicle is the target speed, it is determined that the vehicle is in an abnormal scene of the positioning module; wherein the target scene includes: an abnormal scene of the positioning module.
  • determining whether the vehicle is in a target scenario according to the target data includes: when the predictive planning data is used to indicate the first In the case of the predicted movement state of three obstacles, determine the similarity between the predicted movement state and the target movement state of the third obstacle, and when the similarity is less than an eighth threshold, determine the vehicle In an abnormal scenario of the prediction planning module, the target movement state is the movement state obtained after detecting the third obstacle; when the prediction planning data is used to indicate the planned movement trajectory of the vehicle , determine the feasibility of the planned movement trajectory, and when the feasibility is less than the ninth threshold, determine that the vehicle is in an abnormal scenario of the prediction planning module; where the prediction planning data is used to indicate the planning of the vehicle In the case of a movement trajectory, determine whether there are obstacles on the planned movement trajectory, and if it is determined that there are obstacles on the planned movement trajectory, determine that the vehicle is in an abnormal scenario of the prediction planning module; in the prediction planning Data used to indicate that the autonomous driving system is not Under the condition that the
  • collecting vehicle scene data of the vehicle in the target scene includes: collecting body data of the vehicle within a preset time period, wherein within the preset time period, , the vehicle is in the target scene; the chassis data of the vehicle is collected within the preset time period; the process data and control data generated by the automatic driving system of the vehicle during the preset time period are collected Instruction data: Collect the sensing data of the environment in which the vehicle is located, determined by the image acquisition device and the radar sensor within the preset time period.
  • the method further includes: sending the vehicle scene data to a cloud server, so that the cloud server can The vehicle scene data adjusts the algorithm of the automatic driving system; obtains the adjusted target algorithm from the cloud server, and updates the algorithm of the automatic driving system to the target algorithm.
  • the present disclosure also provides a device for collecting vehicle scene data.
  • the device includes: an acquisition module configured to acquire target data determined by the automatic driving system of the vehicle while the vehicle is driving, wherein, The target data includes at least one of the following: the control instructions of the automatic driving system to the vehicle, the sensing data of the environment in which the vehicle is located, the sensing data determined by the automatic driving system.
  • the positioning data of the vehicle and the prediction planning data predicted by the automatic driving system predicted by the automatic driving system; a determination module configured to determine whether the vehicle is in the target scene according to the target data; an acquisition module configured to determine whether the vehicle is in the target scene In the case of , collect vehicle scene data of the vehicle in the target scene.
  • the present disclosure also provides a computer-readable storage medium.
  • the computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to perform any of the above.
  • the present disclosure also provides an electronic device, the electronic device includes one or more processors; a storage device configured to store one or more programs, when the one or more programs are processed by the one When executed by or multiple processors, the one or more processors are configured to run a program, wherein the program is configured to execute the vehicle scene data collection method described in any of the above technical solutions when running. .
  • the vehicle automatically determines whether it is within the set target. Scenarios trigger data collection. Since the vehicle's ability to recognize scenes is stronger than that of manual recognition, the vehicle can recognize more scenes, and the scenes in which the vehicle is driving are diverse, so it can collect A large amount of real and diverse scene data solves the problem of less vehicle scene data being obtained.
  • Figure 1 is a hardware structure block diagram of a computer terminal of a vehicle scene data collection method according to an embodiment of the present disclosure
  • Figure 2 is a flow chart of a vehicle scene data collection method according to an embodiment of the present disclosure
  • Figure 3 is the hardware architecture of a vehicle-side data collection unit according to an embodiment of the present disclosure
  • Figure 4 is a data collection triggering mechanism diagram (1) according to an embodiment of the present disclosure
  • Figure 5 is a schematic diagram of a data collection triggering scenario according to an embodiment of the present disclosure.
  • Figure 6 is a data collection triggering mechanism diagram (2) according to an embodiment of the present disclosure.
  • FIG. 7 is a structural block diagram of a device for collecting vehicle scene data according to an embodiment of the present disclosure.
  • FIG. 1 is a hardware structure block diagram of a computer terminal for a vehicle scene data collection method according to an embodiment of the present disclosure.
  • the computer terminal may include one or more (one is shown in FIG. 1 ) processors 102 (the processor 102 may include but is not limited to a microprocessor unit (MPU for short) or a programmable logic device. (Programmable logic device, referred to as PLD)) and a memory 104 configured to store data.
  • MPU microprocessor unit
  • PLD programmable logic device
  • the above-mentioned computer terminal may also include a transmission device 106 configured to have a communication function and an input and output device 108.
  • a transmission device 106 configured to have a communication function
  • an input and output device 108 an input and output device 108.
  • FIG. 1 is schematic and does not limit the structure of the above-mentioned computer terminal.
  • a computer terminal may also include more or fewer components than shown in Figure 1, Or a different configuration with equivalent functions or more functions than those shown in FIG. 1 .
  • the memory 104 may be configured to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the vehicle scene data collection method in the embodiment of the present disclosure.
  • the processor 102 runs the computer program stored in the memory 104 , thereby executing various functional applications and data processing, that is, implementing the above method.
  • Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory.
  • the memory 104 may further include memory located remotely relative to the processor 102, and these remote memories may be connected to the computer terminal through a network. Examples of the above-mentioned networks include but are not limited to the Internet, intranets, local area networks, mobile communication networks and combinations thereof.
  • the transmission device 106 is arranged to receive or send data via a network.
  • Specific examples of the above-mentioned network may include a wireless network provided by a communication provider of the computer terminal.
  • the transmission device 106 includes a network adapter (Network Interface Controller, NIC for short), which can be connected to other network devices through a base station to communicate with the Internet.
  • the transmission device 106 may be a radio frequency (Radio Frequency, RF for short) module, which is configured to communicate with the Internet wirelessly.
  • this disclosure proposes a vehicle scene data collection method, which can be used in automatic driving and manual driving respectively.
  • Real-scenario data collection is carried out under driving conditions.
  • the autonomous driving system continues to run.
  • autonomous driving mode instructions are output to control vehicle driving.
  • manual driving mode virtual instructions are output to compare the differences with the actual driver.
  • human and vehicle Difference comparison analysis and set scenes of interest effectively trigger and collect real scene data on the vehicle side.
  • FIG. 1 is a flow chart of a vehicle scene data collection method according to an embodiment of the present disclosure. The process includes the following steps:
  • Step S202 While the vehicle is driving, obtain the target data determined by the automatic driving system of the vehicle, where the target data includes at least one of the following: the control instructions of the automatic driving system for the vehicle, the Sensing data perceived by the autonomous driving system of the environment in which the vehicle is located, positioning data of the vehicle determined by the autonomous driving system, and predictive planning data predicted by the autonomous driving system;
  • control instructions include but are not limited to: control vehicle acceleration, control vehicle braking and deceleration, control vehicle turning, control vehicle U-turn, control vehicle turn on lights and other control instructions.
  • the above-mentioned sensing data includes but is not limited to: sensing the number, type, identification, speed, location, driving status, etc. of obstacles around the vehicle.
  • the above positioning data includes but is not limited to: the position of the vehicle, the moving direction (heading) of the vehicle, etc.
  • the above-mentioned predictive planning data includes but is not limited to: the predicted movement state of the obstacle (for example, the obstacle moves in the x direction at x speed), the planned travel trajectory of the vehicle, and so on.
  • Step S204 Determine whether the vehicle is in a target scene according to the target data
  • the target scenario is the scenario where the tester wants to collect vehicle scene data.
  • the target scenario includes but is not limited to: abnormal driving scenarios, abnormal predictive control instruction scenarios, abnormal sensing module scenarios, preset scenarios customized by the tester, Abnormal scenarios of the positioning module, abnormal scenarios of the prediction and planning module, and scenarios to be manually controlled. These scenarios will be described in detail below and will not be described in detail here.
  • Step S206 When it is determined that the vehicle is in the target scene, collect vehicle scene data of the vehicle in the target scene.
  • collecting vehicle scene data of the vehicle in the target scene can be achieved in the following ways:
  • vehicle scene data includes: body data, chassis data, process data generated by the autonomous driving system, control instruction data, and perception data.
  • Figure 3 is the hardware architecture of a vehicle-side data collection unit according to an embodiment of the present disclosure.
  • Vehicle scene data can be collected through the vehicle-side data collection unit as shown in Figure 3, where the vehicle-side data collection unit includes: Data acquisition triggering calculation unit, data storage unit;
  • the input of the vehicle-side data collection unit includes:
  • the vehicle scene data also needs to be sent to the cloud server, so that the cloud server adjusts the algorithm of the automatic driving system according to the vehicle scene data; obtain The cloud server adjusts the target algorithm, and updates the algorithm of the autonomous driving system to the target algorithm.
  • the vehicle scene data stored in the data storage unit in the vehicle-side data acquisition unit can be uploaded to the cloud server through the communication module as shown in Figure 3.
  • the vehicle automatically determines whether it is in the set target scene, and then triggers data collection. Since the vehicle's ability to identify scenes is stronger than that of manual identification, the vehicle can then identify There are more scenes, and the scenes in which the vehicle is driving are diverse, so a large amount of real and diverse scene data can be collected, thus solving the problem of less vehicle scene data being obtained.
  • determining whether the vehicle is in the target scene according to the target data is implemented by the following methods one to three:
  • Method 1 When it is determined that the vehicle is in the automatic driving mode and the control instruction indicates that the acceleration of the vehicle in the first direction is set to the first acceleration, it is determined that the vehicle is in an abnormal driving scenario, wherein, The first direction is the forward direction of the vehicle, the value of the first acceleration exceeds a first threshold, and the target scene includes: an abnormal driving scene;
  • control instruction indicates that the acceleration of the vehicle in the first direction is set to the first acceleration, it means that the automatic driving system needs to control the vehicle to brake suddenly, that is, the current vehicle is in a sudden extreme state, that is, it is in an abnormal driving scenario. Down.
  • Method 2 After it is determined that the vehicle is in the autonomous driving mode, the control instruction instructs to set the acceleration of the vehicle in the second direction as the second acceleration, and instructs to set the rotation angle change rate of the vehicle's tires as the target rotation angle.
  • change rate it is determined that the vehicle is in an abnormal driving scene, wherein the second acceleration exceeds a second threshold, the target angle change rate exceeds a third threshold, and the second direction is The angle between is the preset angle;
  • control instruction indicates that the acceleration of the vehicle in the second direction is set to the second acceleration and the rotation angle change rate of the vehicle's tires is set to the target angle change rate, it means that the automatic driving needs to control the vehicle to turn sharply. Then the current vehicle is in a sudden extreme state, that is, in an abnormal driving scenario.
  • Method 3 When it is determined that the vehicle is in manual driving mode, determine the target control instruction issued by the target object to the vehicle; compare the target control instruction with the control instruction in the target data, and compare the target control instruction with the control instruction in the target data. If the similarity between the target control instruction and the control instruction is less than the fourth threshold, it is determined that the vehicle is in a predictive control instruction abnormal scene, where the target scene includes: a predictive control instruction abnormal scene.
  • the automatic driving system simulates operation, and the control instructions issued by it do not directly control the vehicle. At this time, the driver's operating behavior and the actual state of the vehicle are different from the control instructions of the automatic driving system. After analysis, when the difference between the two exceeds the set fourth threshold, it means that the vehicle is determined to be in an abnormal scenario of predictive control instructions.
  • the target control instruction is to control the vehicle to turn left, and the control instruction is to control the vehicle to turn right, it means that the similarity between the two is less than the fourth threshold.
  • determining whether the vehicle is in the target scene based on the target data can be determined through the following methods four to nine:
  • Method 4 When it is determined based on the sensing data that the type of the first obstacle sensed by the sensing module of the automatic driving system changes within the first preset time period, it is determined that the vehicle is in an abnormal scene of the sensing module, Among them, the target scenarios include: abnormal sensing module scenarios;
  • the first preset time period may be 1 second, 2 seconds, etc.
  • the first obstacle is a movable obstacle (for example: movable pedestrians or vehicles on the road). If the type of the first obstacle is in the first If a change occurs within a preset period of time (for example, there is a pedestrian at the beginning and a car behind), it means that the type of obstacle perceived changes, that is, the sensing module is abnormal and the vehicle is in an abnormal scene of the sensing module.
  • Method 5 When it is determined based on the sensing data that the identification of the first obstacle changes within the first preset time period, it is determined that the vehicle is in an abnormal scene of the sensing module;
  • the identity of the first obstacle changes within the first preset time period (for example: initially the identity of the first obstacle is vehicle 1, and then changes to vehicle 2), it means that the perceived obstacle The object's identification jumps, that is, the sensing module is abnormal, and the vehicle is in an abnormal scene of the sensing module.
  • Method 6 When it is determined based on the sensing data that the change in the moving speed of the first obstacle within the first preset time period exceeds the fifth threshold, it is determined that the vehicle is in an abnormal scene of the sensing module;
  • Method 7 When it is determined based on the sensing data that the position of the second obstacle sensed by the sensing module changes within the first preset time period, it is determined that the vehicle is in an abnormal scene of the sensing module;
  • the second obstacle is a stationary obstacle (for example, a tree, a house, etc.). If the position of the second obstacle changes within the first preset time period, it means that the position of the stationary obstacle jumps, that is, the perception The module is abnormal and the vehicle is in a scenario where the sensing module is abnormal.
  • Method 8 When it is determined based on the sensing data that the position of the third obstacle sensed by the sensing module overlaps with the position of the vehicle, it is determined that the vehicle is in an abnormal scene of the sensing module;
  • the third obstacle includes the first obstacle and the second obstacle.
  • Method 9 When the sensing data is preset sensing data and the vehicle body data is preset body data, determine that the vehicle is in a preset scene, where the target scene includes: a preset scene.
  • the autonomous driving system will be developed for specific scenarios, so data collection for characteristic scenarios is required.
  • the preset scenes of interest can be configured on the cloud server, and the scenes that need to be collected are sent to the vehicle according to specific needs.
  • the vehicle When the vehicle receives the preset scene requirements, it will identify the scene and trigger data recording. , upload to the cloud server.
  • the cloud server For example, if it is necessary to collect autonomous driving lane changing scene data, it can be sent to the car.
  • the car will then identify the vehicle's lane changing behavior based on the sensing data, then record the overall lane changing process data and upload it to the cloud storage application.
  • Preset scenes can be comprehensively judged based on perception, positioning and high-precision map information, such as: (1) road type (highway/city/park%), shape (straight/curve/U-turn%), number of lanes, traffic lights; (2) ) The number, category, location, speed, etc. of obstacles; (3) The current speed, acceleration, lighting status, etc. of the vehicle.
  • the preset sensing data includes but is not limited to (1) above
  • the preset body Data includes but is not limited to (3) above.
  • determining whether the vehicle is in the target scene based on the target data can be determined through the following methods ten to twelve:
  • Method 10 When it is determined based on the positioning data that the change in the position of the vehicle within the second preset time period exceeds the sixth threshold, it is determined that the vehicle is in an abnormal scene of the positioning module, wherein the target scene includes : Abnormal scenario of the positioning module;
  • the second preset duration may be 1 second or 2 seconds.
  • the change in the vehicle's position within the second preset time period exceeds the sixth threshold (for example, from positioning Beijing to positioning Wuhan), it means that the position of the vehicle itself jumps, that is, the positioning of the autonomous driving system.
  • the module is abnormal and the vehicle is in a positioning module abnormality scenario.
  • Method 11 When it is determined based on the positioning data that the change in the movement direction of the vehicle within the second preset time period exceeds a seventh threshold, it is determined that the vehicle is in an abnormal scene of the positioning module;
  • the seventh threshold for example, a sudden change from forward driving to backward driving
  • the vehicle's own heading jumps that is, the positioning module of the automatic driving system is abnormal, and the vehicle is in a positioning module abnormality scenario.
  • Method 12 When it is determined based on the positioning data that the position of the vehicle has not changed within the second preset time period, but the speed of the vehicle is the target speed, it is determined that the vehicle is in a positioning module abnormality. Scenes;
  • the positioning module of the automatic driving system is abnormal and the vehicle is in a positioning module abnormality scenario.
  • determining whether the vehicle is in the target scene based on the target data can be determined through the following methods thirteen to sixteen:
  • Method 13 In the case where the predicted data is used to indicate the predicted movement state of the third obstacle, determine the similarity between the predicted movement state and the target movement state of the third obstacle, and determine the similarity between the predicted movement state and the target movement state of the third obstacle. If the degree is less than the eighth threshold, it is determined that the vehicle is in an abnormal scenario of the predictive planning module, wherein the target movement state is the movement state obtained after detecting the third obstacle, and the target scenario includes: predictive planning Module abnormality scenarios;
  • the predictive planning module of the autonomous driving system is abnormal, that is, the vehicle is in the predicted state. Planning module abnormal scenarios.
  • Method 14 When the predicted planning data is used to indicate the planned movement trajectory of the vehicle, determine the feasibility of the planned movement trajectory, and when the feasibility is less than the ninth threshold, determine the feasibility of the planned movement trajectory.
  • the vehicle described above is in an abnormal scenario in the predictive planning module;
  • Method 15 When the predicted planning data is used to indicate the planned movement trajectory of the vehicle, determine whether there are obstacles on the planned movement trajectory, and determine whether there are obstacles on the planned movement trajectory. Next, it is determined that the vehicle is in an abnormal scenario of the prediction planning module;
  • Method 16 When the predictive planning data is used to indicate that the automatic driving system cannot control the vehicle, determine that the vehicle is in a scenario to be manually controlled, where the target scenario includes a scenario to be manually controlled.
  • this disclosure proposes a vehicle scene data collection method, which can trigger data collection in both automatic driving and manual driving modes. It is specifically divided into automatic driving anomaly identification, human driving Vehicle difference analysis and identification and interest scene identification, and then upload the data to the cloud server for storage and application.
  • Figure 4 is a diagram (1) of the data collection trigger mechanism according to an embodiment of the present disclosure. As shown in Figure 4, the data collection trigger mechanism is divided into two situations.
  • FIG. 5 is a schematic diagram of a data collection triggering scene according to an embodiment of the present disclosure.
  • Data recording can be triggered based on the longitudinal (equivalent to the first direction in the above embodiment) acceleration exceeding a certain threshold.
  • Data recording is triggered when common sense errors such as overlapping of the obstacle recognition position with the own vehicle occur.
  • Forecast planning data anomaly (equivalent to the planning prediction module abnormality scenario in the above embodiment)
  • autonomous driving systems generally include perception, positioning, and predictive planning modules.
  • functional modules in the autonomous driving system are abnormal, data recording is triggered.
  • Autonomous driving will be developed for specific scenarios, so data collection for characteristic scenarios is required.
  • the scene of interest is configured in the cloud, and the scene to be collected is sent to the vehicle according to specific needs.
  • the vehicle receives the demand for the scene of interest, it identifies the scene, triggers data recording, and uploads it to cloud. For example, if it is necessary to collect autonomous driving lane changing scene data, it can be sent to the car. The car will then identify the vehicle's lane changing behavior based on the sensing data, then record the overall lane changing process data and upload it to the cloud storage application.
  • Figure 6 is a diagram (2) of the data collection triggering mechanism according to an embodiment of the present disclosure. As shown in Figure 3, the data collection triggering mechanism is divided into three situations:
  • the automatic driving system simulates operation, and its control instructions do not directly control the vehicle. At this time, the driver's operating behavior and the actual state of the vehicle are analyzed for differences with the control instructions of the automatic driving system. When the difference between the two exceeds the set When a set threshold is reached (when the difference between the two exceeds the set threshold, it indicates that the vehicle is in an abnormal scenario of predictive control instructions), data recording is triggered.
  • automatic driving anomaly identification includes sensing data anomalies, positioning data anomalies and planning data anomalies. The triggering methods of these three types of anomalies are consistent with those in automatic driving mode.
  • this disclosure designs a trigger mechanism for data collection and storage in both automatic driving mode and manual driving mode.
  • the automatic driving abnormal scene data trigger and the interesting scene trigger are designed to improve the effective collection of data in the automatic driving mode.
  • manual driving mode through the virtual operation of the automatic driving system, based on the collection of abnormal scene data of automatic driving and scenes of interest, the difference scene trigger between humans and vehicles can be added to quickly collect the gap between the automatic driving system and the manual driver. Effectively improve the coverage of vehicle-side scenarios and provide scenario data support for the optimization and development of autonomous driving systems.
  • abnormal scenario data of autonomous driving data collection is triggered in five situations, including abnormal takeover scenarios of autonomous driving, abnormal autonomous driving control instructions, and perception Data anomalies, positioning data anomalies and predictive planning data anomalies can effectively trigger abnormal scenarios from the overall and sub-module levels.
  • a scene identification module of interest is designed, and a cloud scene configuration function is designed, which can remotely deliver scenes of interest in the cloud and quickly realize data collection of specific scenes.
  • the method according to the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is Better implementation.
  • the technical solution of the present disclosure can be embodied in the form of a software product in essence or that contributes to the existing technology.
  • the computer software product is stored in a storage medium (such as ROM/RAM, disk, CD), including several instructions to cause a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of various embodiments of the present disclosure.
  • This embodiment also provides a device for collecting vehicle scene data.
  • the device is configured to implement the above embodiments and preferred implementations. What has already been explained will not be described again.
  • the term "module” may be a combination of software and/or hardware that implements a predetermined function.
  • the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.
  • Figure 7 is a structural block diagram of a device for collecting vehicle scene data according to an embodiment of the present disclosure.
  • the device includes:
  • the acquisition module 72 is configured to acquire target data determined by the automatic driving system of the vehicle while the vehicle is driving, wherein the target data includes at least one of the following: control instructions for the vehicle by the automatic driving system. , the sensing data perceived by the automatic driving system of the environment in which the vehicle is located, the positioning data of the vehicle determined by the automatic driving system, and the predictive planning data predicted by the automatic driving system;
  • Determining module 74 configured to determine whether the vehicle is in a target scene according to the target data
  • the collection module 76 is configured to collect vehicle scene data of the vehicle in the target scene when it is determined that the vehicle is in the target scene.
  • the vehicle when the vehicle is driving, the vehicle automatically determines whether the setting In the target scene, data collection is triggered. Since the vehicle's ability to recognize scenes is stronger than that of manual recognition, the vehicle can recognize more scenes, and the scenes in which the vehicle is driving are diverse, so it can A large amount of real and diverse scene data is collected, thus solving the problem of less vehicle scene data being obtained.
  • the determining module is configured to determine whether the vehicle is in a target scene in the following manner: when determining that the vehicle is in an autonomous driving mode, And when the control instruction indicates that the acceleration of the vehicle in the first direction is set to the first acceleration, it is determined that the vehicle is in an abnormal driving scene, wherein the first direction is the forward direction of the vehicle, The value of the first acceleration exceeds a first threshold; or when it is determined that the vehicle is in the automatic driving mode, the control instruction instructs to set the acceleration of the vehicle in the second direction to the second acceleration and instructs the vehicle to When the tire's rotation angle change rate is set to the target rotation angle change rate, it is determined that the vehicle is in an abnormal driving scene, wherein the second acceleration exceeds a second threshold, the target rotation angle change rate exceeds a third threshold, and the The angle between the second direction and the second direction is a preset angle; when it is determined that the vehicle is in manual driving mode, determine the target control instruction issued by the target object to the vehicle; control the target The instruction is compared
  • the determining module is configured to determine whether the vehicle is in the target scene in the following manner: when the target data includes the sensing data, the automatic When the type of the first obstacle sensed by the sensing module of the driving system changes within the first preset time period, it is determined that the vehicle is in an abnormal scene of the sensing module; or when the first obstacle is determined based on the sensing data
  • the identification changes within the first preset time period it is determined that the vehicle is in an abnormal scene of the sensing module; or when it is determined based on the sensing data that the change in the moving speed of the first obstacle within the first preset time period exceeds
  • the fifth threshold it is determined that the vehicle is in an abnormal scene of the sensing module; or it is determined based on the sensing data that the position of the second obstacle sensed by the sensing module changes within the first preset time period.
  • the vehicle is in a sensing module abnormality scene; when it is determined according to the sensing data that the position of the third obstacle sensed by the sensing module overlaps with the position of the vehicle, it is determined that the vehicle is in a sensing module abnormality scene. scene; described When the sensing data is preset sensing data and the vehicle body data is preset body data, it is determined that the vehicle is in a preset scene; wherein the target scene includes: an abnormal scene of the sensing module, and the Preset scenes.
  • the determination module is configured to determine whether the vehicle is in the target scene in the following manner: when the target data includes the positioning data, the vehicle is determined according to the positioning data. If the change in the position of the vehicle exceeds the sixth threshold within the second preset time period, it is determined that the vehicle is in an abnormal scene of the positioning module; or it is determined based on the positioning data that the moving direction of the vehicle is in the second preset time period.
  • the seventh threshold it is determined that the vehicle is in an abnormal scene of the positioning module; or it is determined based on the positioning data that the position of the vehicle has not changed within the second preset duration, However, when the speed of the vehicle is the target speed, it is determined that the vehicle is in an abnormal scene of the positioning module; wherein the target scene includes: an abnormal scene of the positioning module.
  • the determination module is configured to determine whether the vehicle is in a target scenario in the following manner: when the target data includes the predictive planning data, the predictive planning data is set to indicate In the case of the predicted movement state of the third obstacle, determine the similarity between the predicted movement state and the target movement state of the third obstacle, and when the similarity is less than an eighth threshold, determine the The vehicle is in an abnormal scenario of the prediction planning module, wherein the target movement state is the movement state obtained after detecting the third obstacle; or the prediction planning data is set to indicate the planned movement trajectory of the vehicle.
  • determine the feasibility of the planned movement trajectory and when the feasibility is less than the ninth threshold, determine that the vehicle is in an abnormal scenario of the predictive planning module; or when the predictive planning data is set to indicate the In the case of the planned movement trajectory of the vehicle, determine whether there are obstacles on the planned movement trajectory, and if it is determined that there are obstacles on the planned movement trajectory, determine that the vehicle is in an abnormal scenario of the prediction planning module; or When the predictive planning data is set to indicate that the automatic driving system cannot control the vehicle, it is determined that the vehicle is in a scenario to be manually controlled; wherein the target scenario includes: an abnormal scenario of the predictive planning module, and the The scene is to be manually controlled.
  • the determination module is configured to collect vehicle scene data of the vehicle in the target scene in the following manner: collect vehicle body data of the vehicle within a preset time period, wherein, in the During the preset time period, the vehicle is in the target scene; the chassis data of the vehicle during the preset time period is collected; and the chassis data of the vehicle during the preset time period is collected.
  • the process data and control instruction data generated by the automatic driving system are collected; the perception data of the environment of the vehicle determined by the image acquisition device and the radar sensor within the preset time period are collected.
  • the above device further includes: a processing module configured to, after collecting vehicle scene data of the vehicle in the target scene, send the vehicle scene data to the cloud server so that the The cloud server adjusts the algorithm of the automatic driving system according to the vehicle scene data; obtains the adjusted target algorithm from the cloud server, and updates the algorithm of the automatic driving system to the target algorithm.
  • a processing module configured to, after collecting vehicle scene data of the vehicle in the target scene, send the vehicle scene data to the cloud server so that the The cloud server adjusts the algorithm of the automatic driving system according to the vehicle scene data; obtains the adjusted target algorithm from the cloud server, and updates the algorithm of the automatic driving system to the target algorithm.
  • Embodiments of the present disclosure also provide a computer-readable storage medium that stores a computer program, wherein the computer program is configured to execute the steps in any of the above method embodiments when running.
  • the above-mentioned storage medium may be configured to store a computer program configured to perform the following steps:
  • the target data includes at least one of the following: the control instruction of the automatic driving system to the vehicle, the automatic driving system Sensing data perceived by the driving system of the environment in which the vehicle is located, positioning data of the vehicle determined by the automatic driving system, and predictive planning data predicted by the automatic driving system;
  • the computer-readable storage medium may include but is not limited to: U disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM) , mobile hard disk, magnetic disk or optical disk and other media that can store computer programs.
  • ROM read-only memory
  • RAM random access memory
  • mobile hard disk magnetic disk or optical disk and other media that can store computer programs.
  • the above-mentioned storage medium of the present disclosure may be a computer-readable signal medium or or a computer-readable storage medium or any combination of the above two.
  • the computer-readable storage medium may be, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any combination thereof. More specific examples of computer readable storage media may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard drive, random access memory (RAM), read only memory (ROM), removable Programmd read-only memory (EPROM or flash memory), fiber optics, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • a computer-readable storage medium may be any tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device.
  • a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above.
  • a computer-readable signal medium may also be any storage medium other than computer-readable storage media that can transmit, propagate, or transport a program configured for use by or in connection with an instruction execution system, apparatus, or device.
  • the program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.
  • Embodiments of the present disclosure also provide an electronic device, including a memory and a processor.
  • a computer program is stored in the memory, and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.
  • the above-mentioned processor may be configured to perform the following steps through a computer program:
  • the target data includes at least one of the following: the control instruction of the automatic driving system to the vehicle, the automatic driving system Sensing data perceived by the driving system of the environment in which the vehicle is located, positioning data of the vehicle determined by the automatic driving system, and predictive planning data predicted by the automatic driving system;
  • the above-mentioned electronic device may further include a transmission device and an input-output device, wherein the transmission device is connected to the above-mentioned processor, and the input-output device is connected to the above-mentioned processor.
  • modules or steps of the present disclosure can be implemented using general-purpose computing devices, and they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices. They may be implemented in program code executable by a computing device, such that they may be stored in a storage device for execution by the computing device, and in some cases may be executed in a sequence different from that shown herein. Or the described steps can be implemented by making them into individual integrated circuit modules respectively, or by making multiple modules or steps among them into a single integrated circuit module. As such, the present disclosure is not limited to any specific combination of hardware and software.
  • the vehicle scene data collection method and device provided by the embodiments of the present disclosure can be applied to the process of driving the vehicle, and the vehicle automatically determines whether it is in the set target scene, thereby triggering data collection. Since the vehicle's ability to recognize scenes is better than manual The ability to recognize scenes is stronger, and the vehicle can recognize more scenes, and the scenes in which the vehicle is driving are diverse, and a large amount of real and diverse scene data can be collected, thus solving the problem of obtaining vehicle The problem of less scene data.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • Traffic Control Systems (AREA)
  • Control Of Driving Devices And Active Controlling Of Vehicle (AREA)

Abstract

Des modes de réalisation de la présente divulgation concernent un procédé et un appareil d'acquisition de données de scène de véhicule, un support de stockage et un dispositif électronique. Le procédé comprend les étapes suivantes : dans le processus d'exécution de véhicule, obtenir des données cibles déterminées par un système de conduite automatique du véhicule, les données cibles comprenant au moins l'un des éléments suivants : une instruction de commande du système de conduite automatique sur le véhicule, des données de détection détectées par le système de conduite automatique pour l'environnement où se trouve le véhicule, des données de positionnement du véhicule déterminées par le système de conduite automatique, et des données de planification de prédiction prédites par le système de conduite automatique ; déterminer si le véhicule se trouve dans une scène cible selon les données cibles ; et dans la condition selon laquelle il est déterminé que le véhicule se trouve dans la scène cible, collecter des données de scène de véhicule du véhicule dans la scène cible.
PCT/CN2023/092610 2022-05-27 2023-05-06 Procédé et appareil d'acquisition de données de scène de véhicule, support de stockage et dispositif électronique WO2023226733A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210590860.X 2022-05-27
CN202210590860.XA CN115019267A (zh) 2022-05-27 2022-05-27 车辆场景数据的采集方法、装置、存储介质及电子设备

Publications (2)

Publication Number Publication Date
WO2023226733A1 true WO2023226733A1 (fr) 2023-11-30
WO2023226733A9 WO2023226733A9 (fr) 2024-01-04

Family

ID=83071249

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/092610 WO2023226733A1 (fr) 2022-05-27 2023-05-06 Procédé et appareil d'acquisition de données de scène de véhicule, support de stockage et dispositif électronique

Country Status (2)

Country Link
CN (1) CN115019267A (fr)
WO (1) WO2023226733A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115019267A (zh) * 2022-05-27 2022-09-06 中国第一汽车股份有限公司 车辆场景数据的采集方法、装置、存储介质及电子设备
CN116664964B (zh) * 2023-07-31 2023-10-20 福思(杭州)智能科技有限公司 数据筛选方法、装置、车载设备和存储介质

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020053170A1 (fr) * 2018-09-10 2020-03-19 Volkswagen Aktiengesellschaft Procédé de génération d'une collecte d'informations sur des scénarios de roulement d'au moins un véhicule, et véhicule, dispositif et système formé de ceux-ci
CN111399481A (zh) * 2020-03-30 2020-07-10 东风汽车集团有限公司 一种自动驾驶场景信息收集及远程升级方法及系统
CN112233273A (zh) * 2020-10-10 2021-01-15 苏州智加科技有限公司 数据传输方法、装置、车载设备及存储介质
CN112562114A (zh) * 2020-12-18 2021-03-26 北京百度网讯科技有限公司 数据采集的方法及装置
CN113743356A (zh) * 2021-09-15 2021-12-03 东软睿驰汽车技术（沈阳）有限公司 数据的采集方法、装置和电子设备
CN113761306A (zh) * 2020-09-30 2021-12-07 北京京东乾石科技有限公司 一种车端数据处理方法及装置
CN115019267A (zh) * 2022-05-27 2022-09-06 中国第一汽车股份有限公司 车辆场景数据的采集方法、装置、存储介质及电子设备
CN115923820A (zh) * 2023-01-19 2023-04-07 蔚来汽车科技(安徽)有限公司 用于车辆的自动驾驶系统的场景数据收集方法和装置

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020053170A1 (fr) * 2018-09-10 2020-03-19 Volkswagen Aktiengesellschaft Procédé de génération d'une collecte d'informations sur des scénarios de roulement d'au moins un véhicule, et véhicule, dispositif et système formé de ceux-ci
CN111399481A (zh) * 2020-03-30 2020-07-10 东风汽车集团有限公司 一种自动驾驶场景信息收集及远程升级方法及系统
CN113761306A (zh) * 2020-09-30 2021-12-07 北京京东乾石科技有限公司 一种车端数据处理方法及装置
CN112233273A (zh) * 2020-10-10 2021-01-15 苏州智加科技有限公司 数据传输方法、装置、车载设备及存储介质
CN112562114A (zh) * 2020-12-18 2021-03-26 北京百度网讯科技有限公司 数据采集的方法及装置
CN113743356A (zh) * 2021-09-15 2021-12-03 东软睿驰汽车技术（沈阳）有限公司 数据的采集方法、装置和电子设备
CN115019267A (zh) * 2022-05-27 2022-09-06 中国第一汽车股份有限公司 车辆场景数据的采集方法、装置、存储介质及电子设备
CN115923820A (zh) * 2023-01-19 2023-04-07 蔚来汽车科技(安徽)有限公司 用于车辆的自动驾驶系统的场景数据收集方法和装置

Also Published As

Publication number Publication date
WO2023226733A9 (fr) 2024-01-04
CN115019267A (zh) 2022-09-06

Similar Documents

Publication Publication Date Title
WO2023226733A1 (fr) Procédé et appareil d'acquisition de données de scène de véhicule, support de stockage et dispositif électronique
JP7030044B2 (ja) 自律走行車（ａｄｖ）に対して車両とクラウド間のリアルタイム交通地図を構築するためのシステム
US11427211B2 (en) Methods for handling sensor failures in autonomous driving vehicles
US10807599B2 (en) Driving scenario based lane guidelines for path planning of autonomous driving vehicles
US20190318267A1 (en) System and method for training a machine learning model deployed on a simulation platform
US20180374360A1 (en) Traffic prediction based on map images for autonomous driving
JP2023153943A (ja) 検出された障壁に基づく車両のナビゲーション
KR20200126935A (ko) 센서 고장에 대한 자율 주행 차량 대응방안을 구현하기 위한 시스템 및 방법
US20190317520A1 (en) Learning based speed planner for autonomous driving vehicles
US20210200217A1 (en) Systems and Methods for Matching an Autonomous Vehicle to a Rider
CN112714729A (zh) 车辆的路径规划方法以及车辆的路径规划装置
US10860868B2 (en) Lane post-processing in an autonomous driving vehicle
US10795357B2 (en) Method and device for controlling vehicle and autonomous driving vehicle
CN113228135B (zh) 一种盲区图像获取方法及相关终端装置
US11267464B2 (en) System and method to distinguish between moving and static objects
CN110888434A (zh) 自动驾驶方法、装置、计算机设备和计算机可读存储介质
WO2023066146A1 (fr) Procédé et appareil de commande d'évitement de véhicule, et dispositif électronique et support d'enregistrement
CN117130298A (zh) 用于评估自动驾驶系统的方法、装置及存储介质
US20230256994A1 (en) Assessing relative autonomous vehicle performance via evaluation of other road users
EP4082856A2 (fr) Évaluateur basé sur l'apprentissage e2e pour un véhicule de conduite autonome
CN112829762A (zh) 一种车辆行驶速度生成方法以及相关设备
JP2020149323A (ja) 情報処理装置及び情報処理装置を備える自動走行制御システム
WO2022148068A1 (fr) Procédé de détection de véhicule et appareil de détection de véhicule
US20230042001A1 (en) Weighted planning trajectory profiling method for autonomous vehicle
US11787407B2 (en) System and method for sensing vehicles and street

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: 23810806

Country of ref document: EP

Kind code of ref document: A1