Intelligent driving software testing method, system and equipment and storage medium

Abstract

The invention relates to an intelligent driving software testing method, system and device and a storage medium, and the method comprises the steps: obtaining current environment data matched with a current state based on the current state of a simulation vehicle and labels associated with multi-frame environment data, and enabling each frame of environment data to be associated with at least two labels, the at least two labels respectively reflect the collection time and position of each frame of environment data, and the current state comprises the current position and the current time; performing reconstruction based on the current environment data to obtain reconstructed environment data; and controlling the simulation vehicle according to the reconstruction environment data so as to update the current state of the simulation vehicle. According to the embodiment of the invention, the accuracy of the test result of the intelligent driving software can be improved.

Description

Technical Field

[0001] The present invention relates to the field of software testing technology, and in particular to an intelligent driving software testing method, system, device and storage medium. Background Art

[0002] Before the intelligent driving software is applied, it is necessary to test whether the intelligent driving software can perform its functions, whether it can run safely, or whether it is still stable and reliable after long-term operation or repeated use. With the improvement of the automation level of intelligent driving, the scenarios faced in the intelligent driving process gradually tend to be extreme, marginal and difficult to reproduce, so the testing requirements for intelligent driving software are getting higher and higher.

[0003] In the related technology, a simulation test system is used to test the software. Specifically, the scene data is generated based on the road data collected by the real vehicle, and then the scene data is fed back to the simulation test system. The simulation test system reproduces the scene through the scene data, and the intelligent driving software performs driving response based on the reproduced scene to complete the test of the intelligent driving software. In the process of the intelligent driving software performing driving response based on the reproduced scene, the driving response is performed based on the scene data at the same time step.

[0004] However, in the actual simulation test process, it is difficult to achieve complete consistency between the control of the simulated vehicle and the real vehicle. Due to the difference in control between the simulated vehicle and the real vehicle, the accuracy of the test results when the above-mentioned related technologies are used to test the intelligent driving software is not high. Summary of the invention

[0005] One of the purposes of the present invention is to provide an intelligent driving software testing method to solve the problem of low accuracy of the test results of intelligent driving software in the prior art; the second purpose is to provide an intelligent driving software testing device; the third purpose is to provide an electronic device.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A method for testing intelligent driving software, comprising:

[0008] Based on the current state of the simulated vehicle and the tags associated with multiple frames of environmental data, current environmental data matching the current state is obtained, wherein each frame of the environmental data is associated with at least two tags, and the at least two tags respectively reflect the time and location of acquisition of each frame of the environmental data, and the current state includes the current location and the current time;

[0009] Reconstructing based on the current environment data to obtain reconstructed environment data;

[0010] The simulated vehicle is controlled according to the reconstructed environment data to update the current state of the simulated vehicle.

[0011] According to the above technical means, by formulating multiple labels for the recycled road sampling environmental data, when testing the intelligent driving software, the current state of the simulated vehicle is matched with multiple labels associated with the environmental data, and the road sampling environmental data that is closer to the real-time state of the simulated vehicle can be matched. The driving scene of the simulated vehicle can be reconstructed based on the road sampling environmental data, and the intelligent driving software can plan and control the path according to the accurate driving scene to control the driving of the simulated vehicle. Since the driving scene has a high degree of matching with the current state of the simulated vehicle, the control result of the simulated vehicle can accurately reflect the test results reflected by the decision of the intelligent driving software in the corresponding driving scene, and the test results of the intelligent driving software are highly accurate.

[0012] Further, based on the current state of the simulated vehicle and the tags associated with the multiple frames of environmental data, current environmental data matching the current state is obtained, including:

[0013] Finding a target position with the shortest distance from the current position of the simulated vehicle from the positions associated with the multiple frames of environmental data;

[0014] Based on the environmental data associated with the target location and the current time, current environmental data matching the current state is obtained.

[0015] According to the above technical means, the test of intelligent driving software is more flexible. During the test, the driving path of the simulated vehicle can be adjusted according to the environmental data (a more reasonable driving path can be planned), and it does not need to be completely consistent with the actual driving path of the real vehicle. And when the actual driving paths of the simulated vehicle and the real vehicle are inconsistent, it is also possible to find the target position with the smallest distance from the current position of the simulated vehicle from the positions associated with the multi-frame environmental data, and find the current environmental data closest to or most matching the current environment of the simulated vehicle from the environmental data associated with the target position in combination with the current time, so as to ensure the accuracy of the reconstructed scene data obtained by the intelligent driving software, and then ensure the accuracy of the test results.

[0016] Further, based on the environmental data associated with the target location and the current time, obtaining current environmental data matching the current state includes:

[0017] Finding a target time having a minimum time interval with the current time from the environmental data associated with the target location;

[0018] Based on the target time, current environment data matching the current state is determined.

[0019] According to the above technical means, the test of intelligent driving software is more flexible. During the test, the driving path of the simulated vehicle can be adjusted according to the environmental data (a more reasonable driving path can be planned), and it does not need to be completely consistent with the actual driving path of the real vehicle. And when the actual driving paths of the simulated vehicle and the real vehicle are inconsistent, it is also possible to find the target position with the smallest distance from the current position of the simulated vehicle from the positions associated with the multi-frame environmental data, and find the target time with the smallest time interval with the current time from the environmental data associated with the target position, and then obtain the current environmental data closest to or most matching the current environment of the simulated vehicle based on the environmental data associated with the target time, so as to ensure the accuracy of the reconstructed scene data obtained by the intelligent driving software, and then ensure the accuracy of the test results.

[0020] Further, the obtaining of current environmental data matching the current state based on the current state of the simulated vehicle and the labels associated with the multi-frame environmental data includes:

[0021] Based on the current gear position of the simulated vehicle, multiple frames of environmental data identical to the current gear position are obtained, wherein the current state also includes the current gear position, and the at least two tags also reflect the vehicle gear position when each frame of the environmental data is collected;

[0022] Based on the current time and current position of the simulated vehicle, and the collected time and position corresponding to the multiple frames of environmental data, current environmental data matching the current state is obtained.

[0023] According to the above technical means, by filtering the environmental data that does not match the gear through the current gear of the vehicle, the amount of matching data can be reduced. At the same time, matching the environmental data based on the same position passed during the reversing and forward processes can also be avoided.

[0024] Furthermore, the environmental data includes static targets and dynamic targets, and the reconstructing based on the current environmental data to obtain the reconstructed environmental data includes:

[0025] Based on the current environment data, a static target is obtained;

[0026] Determining a dynamic target based on current environmental data obtained by the simulated vehicle in various states;

[0027] Based on the simulated radar position and the simulated radar parameters, simulating and calculating the radar detection data obtained when the simulated radar detects the static target and the dynamic target in the current state;

[0028] Based on the reconstruction of the static target, the dynamic target and the radar detection data, reconstructed environment data is obtained.

[0029] According to the above technical means, on the one hand, the radar detection data is calculated based on the corrected static targets and dynamic targets, and the radar data deviation caused by the deviation between the real vehicle and the simulated vehicle is corrected; on the other hand, based on the simulated radar position and simulated radar parameters, the radar detection data obtained by detecting these static targets and dynamic targets is simulated and calculated. It does not require that the simulated radar position and simulated radar parameters must be completely consistent with the radar position and radar parameters on the real vehicle. It can be applied to different radars and eliminates equipment deviation.

[0030] Further, the reconstructing based on the current state of the simulated vehicle to obtain the reconstructed environment data includes:

[0031] Based on the body coordinate system of the simulated vehicle, correcting the current environment data;

[0032] Reconstruction is performed based on the corrected current environment data to obtain reconstructed environment data.

[0033] Further, the correcting the current environment data based on the body coordinates of the simulated vehicle includes:

[0034] Determine a first conversion matrix between the body coordinates of the actual vehicle and the preset coordinates according to the driving information of the actual vehicle when the current environmental data is collected;

[0035] Based on the first conversion matrix, converting the current environment data to the preset coordinates;

[0036] Determine a second conversion matrix between the body coordinates of the simulated vehicle and the preset coordinates according to the current driving information of the simulated vehicle;

[0037] Based on the second conversion matrix, the environmental data in the preset coordinate system is converted to the body coordinate system of the simulated vehicle to obtain corrected current environmental data;

[0038] The driving information includes at least one of a vehicle position, a vehicle heading angle and a vehicle speed.

[0039] According to the above-mentioned technical means, before reconstructing the environmental data of the simulated vehicle, the current environmental data is first converted to the body coordinates of the simulated vehicle to adapt to the position, heading, etc. of the simulated vehicle and the real vehicle. When there is a deviation, the simulated vehicle can obtain accurate environmental data under its own coordinates, which can avoid misjudgment or error of the intelligent driving software and improve the accuracy of the intelligent driving software test results.

[0040] Furthermore, before obtaining the current environmental data matching the current state based on the current state of the simulated vehicle and the tags associated with the multi-frame environmental data, the method further includes:

[0041] Initial road data collection is obtained, and the collection information of the initial road data is used as a label, and the label and the initial road data are associated. The collection information of each frame of environmental data includes the time and location of collection.

[0042] According to the above technical means, the road sampling data is fed back to facilitate the test system to process based on the original data and obtain the required label information.

[0043] Further, controlling the simulated vehicle according to the reconstructed environment data to update the current state of the simulated vehicle includes:

[0044] Obtaining the simulated vehicle control instruction according to the reconstructed environment data;

[0045] According to the control instruction, the simulated vehicle is controlled to update the current state of the simulated vehicle.

[0046] According to the above technical means, the simulated vehicle is controlled according to the reconstructed environmental data to achieve a closed test loop.

[0047] Further, obtaining the simulated vehicle control instruction according to the reconstructed environment data includes:

[0048] updating the to-be-traveled information of the simulated vehicle according to the reconstructed environment data, wherein the to-be-traveled information includes one or more of a trajectory point set, a vehicle speed, an acceleration, and a gear position;

[0049] Determining a travel distance of the simulated vehicle after a preset time step according to the vehicle speed and the acceleration;

[0050] Determine a next target trajectory point according to the current trajectory point where the current position of the simulated vehicle is located and the travel distance;

[0051] Determining a lateral change, a longitudinal change, and a heading angle change of the simulated vehicle according to the next target trajectory point and the current trajectory point;

[0052] A control instruction of the simulated vehicle is obtained according to the lateral variation, the longitudinal variation and the heading angle variation, and the control instruction is used to control the heading and speed of the simulated vehicle.

[0053] According to the above-mentioned technical means, after obtaining the reconstructed environmental data, the simulated vehicle is controlled by considering the most perfect driving state of the vehicle, which can avoid the influence of some additional factors on the control, ensure that the control of the simulated vehicle is more matched with the decision response of the intelligent driving software, better reflect the test results of the intelligent driving software, and improve the reliability of the test results.

[0054] An intelligent driving software testing system, the testing system comprising:

[0055] A data matching module, for obtaining current environmental data matching the current state based on the current state of the simulated vehicle and tags associated with multiple frames of environmental data, wherein each frame of the environmental data is associated with at least two tags, and the at least two tags respectively reflect the time and location of the collection of each frame of the environmental data, and the current state includes a current location and a current time;

[0056] A scene construction module, used for reconstructing based on the current environment data to obtain reconstructed environment data;

[0057] An intelligent driving module is used to determine the driving information of the simulated vehicle according to the reconstructed environment data and the current state.

[0058] The control module is used to update the current state of the simulated vehicle according to the to-be-traveled information.

[0059] An electronic device, comprising: a memory, a processor;

[0060] The memory is used to store computer programs / instructions; the processor is used to implement the above method according to the computer programs / instructions stored in the memory.

[0061] A computer-readable storage medium stores a computer program / instruction, and the computer program / instruction is used to implement the above method when executed by a processor.

[0062] A computer program product, comprising a computer program / instruction, wherein the computer program / instruction is used to implement the above method when executed by a processor.

[0063] Beneficial effects of the present invention:

[0064] (1) In the embodiment of the present application, multiple tags are prepared to associate the recycled road environment data. When testing the intelligent driving software, the multiple tags associated with the environmental data based on the current state of the simulated vehicle are matched to match the road environment data that is closer to the real-time state of the simulated vehicle. The driving scene of the simulated vehicle is reconstructed based on the road environment data. The intelligent driving software can plan and control the path according to the accurate driving scene to control the driving of the simulated vehicle. Since the driving scene has a high degree of match with the current state of the simulated vehicle, the control result of the simulated vehicle can accurately reflect the test results reflected by the decision of the intelligent driving software in the corresponding driving scene, and the test results of the intelligent driving software are highly accurate;

[0065] (2) The embodiments of the present application are suitable for closed-loop test scenarios of simulated vehicles, and for scenarios where it is difficult to achieve complete consistency in the control of the simulated vehicle and the real vehicle, the most suitable environmental data is obtained through time and space synchronization, and the static targets and dynamic targets in the environmental data are corrected. Based on the static targets and dynamic targets, the detection scenario of the simulated radar is simulated to obtain radar detection data, and then the driving scenario of the simulated vehicle is reconstructed based on these data. The deviation between the simulated vehicle and the real vehicle can be corrected. The intelligent test software performs testing based on the driving scenario to improve the reliability and accuracy of the results. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A schematic diagram of a vehicle driving scenario provided in an embodiment of the present application;

[0067] Figure 2 A flowchart of an intelligent driving software testing method provided in one embodiment of the present application;

[0068] Figure 3 A flowchart of an intelligent driving software testing method provided by another embodiment of the present application;

[0069] Figure 4 A schematic diagram of a vehicle driving scenario provided by another embodiment of the present application;

[0070] Figure 5 A flowchart of an intelligent driving software testing method provided by another embodiment of the present application;

[0071] Figure 6 A flowchart of an intelligent driving software testing method provided by another embodiment of the present application;

[0072] Figure 7 A flowchart of an intelligent driving software testing method provided by another embodiment of the present application;

[0073] Figure 8 A schematic diagram of a vehicle driving trajectory provided in an embodiment of the present application;

[0074] Fig. 9 A schematic diagram of the structure of an electronic device provided in this application. DETAILED DESCRIPTION

[0075] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0076] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0077] Before the intelligent driving software is applied, it is necessary to test whether the intelligent driving software can perform its functions, whether it can run safely, or whether it is still stable and reliable after long-term operation or repeated use. With the improvement of the automation level of intelligent driving, the scenarios faced in the intelligent driving process gradually tend to be extreme, marginal and difficult to reproduce, so the testing requirements for intelligent driving software are getting higher and higher.

[0078] In the related technology, a simulation test system is used to test the software. Specifically, the scene data is generated based on the road data collected by the real vehicle, and then the scene data is fed back to the simulation test system. The simulation test system reproduces the scene through the scene data, and the intelligent driving software performs driving response based on the reproduced scene to complete the test of the intelligent driving software. In the process of the intelligent driving software performing driving response based on the reproduced scene, the driving response is performed based on the scene data at the same time step.

[0079] However, in the actual simulation test process, it is difficult to achieve complete consistency between the control of the simulated vehicle and the real vehicle. Due to the differences in the control of the simulated vehicle and the real vehicle (such as time and space differences), when the above-mentioned related technologies are used to test the intelligent driving software, the detection results are not accurate.

[0080] For example, in a simulation test scenario, Figure 1 As shown, the real vehicle collects road data and feeds the road data back to the simulation test system. The simulation test system reproduces the scene, and the intelligent driving software controls the simulated vehicle in the reproduced scene. If at time t1, the real vehicle collects the road data of point A, since the control of the simulated vehicle is difficult to be completely consistent with the control of the real vehicle, the speed of the simulated vehicle may be slower than the speed of the real vehicle. At time t1, the simulated vehicle may arrive at point B instead of point A. In the related art, when the simulated vehicle travels at time t1, the road data of point A is used as a reference, and the driving software identifies based on the road data of point A, and then controls the response of the simulated vehicle. However, the simulated vehicle actually arrives at point B, and the scene corresponding to point B is different from point A, so the test scene of the driving software is actually different from the scene of the real vehicle. The response of the driving software to control the simulated vehicle cannot reflect the test results of the vehicle at point A, resulting in inaccurate test results of the driving software.

[0081] Based on this, an embodiment of the present application provides a method for testing intelligent driving software. First, the actual road data collected by the real vehicle is used as the original data. Then, based on the position and time of the simulated vehicle, better environmental data that is closer to the position and time is selected from the original data. The driving scene of the simulated vehicle is reconstructed based on the better environmental data. The intelligent driving software controls the simulated vehicle based on the driving scene, and the test results reflected are more accurate.

[0082] For example, continue to refer to Figure 1 , the real vehicle collects road data, and feeds the road data back to the simulation test system. The simulation test system reproduces the scene, and the intelligent driving software controls the simulated vehicle in the reproduced scene. If at time t1, the real vehicle collects the road data of point A, because the control of the simulated vehicle is difficult to be completely consistent with the control of the real vehicle, the speed of the simulated vehicle may be slower than the speed of the real vehicle. At time t1, the simulated vehicle may arrive at point B instead of point A. In the embodiment of the present application, based on the position of the simulated vehicle is point B, then based on the position of the simulated vehicle and time t1, the optimal road data (for example, the road data corresponding to point B, or the road data near point B and close in time) is selected from the road data to reproduce the scene, which is closer to the scene of the real vehicle. The intelligent driving software controls the simulated vehicle based on the road data, and the test results are more accurate.

[0083] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0084] Figure 2 FIG. 1 is a flow chart of a method for testing intelligent driving software provided by an embodiment of the present application. Figure 2 As shown, with the electronic device as the execution subject, the electronic device includes an intelligent software testing system, and the method of this embodiment may include the following steps:

[0085] S201, based on the current state of the simulated vehicle and the tags associated with the multiple frames of environmental data, obtaining current environmental data matching the current state;

[0086] In this step, each frame of environmental data is associated with at least two tags, which respectively reflect the time and location of each frame of environmental data collection (the location of the real vehicle where the data is collected). The current state of the simulated vehicle includes the current location and current time of the simulated vehicle.

[0087] As an example, before this step, the following steps are also included:

[0088] Initial road collection data is obtained, and collection information of the initial road collection data is used as a label, and the label and the initial road collection data are associated.

[0089] The initial road data recorded by the real vehicle is fed back into the simulation test system. Before the test, the environmental data in the data is parsed and extracted based on the road data recorded by the real vehicle. The extracted environmental data is the perception information around the vehicle at that moment, including static targets, dynamic targets, ultrasonic radar detection distance, etc.

[0090] In this embodiment, the test system prepares multiple tags associated with environmental data, such as time, location, and gear position of the vehicle, etc. Therefore, when extracting environmental data, it also includes extracting the timestamp, vehicle location, vehicle gear position and other information corresponding to each frame of environmental data, associating the extracted environmental data with the corresponding tags, and obtaining environmental data associated with multiple tags, which is stored in the storage area.

[0091] During the intelligent driving software testing process, multiple frames of environmental data are obtained from the storage area, and then the optimal environmental data is matched based on the current state of the simulated vehicle and the matching degree of the label, as the current environmental data collected by the simulated vehicle in the current state in the simulated environment.

[0092] It should be noted that the test results of the intelligent driving software are accurate only when the simulated vehicle is intelligently driven based on the environmental data corresponding to the same time and position of the real vehicle. Therefore, in the embodiment of the present application, at least based on the consideration of time and space matching, the current environmental data of the simulated vehicle is obtained and the scene is reconstructed, so that the test results of the intelligent driving software in the scene are more accurate.

[0093] As an example, based on the current state of the simulated vehicle and the labels associated with the multi-frame environmental data, current environmental data matching the current state is obtained, which may be specifically:

[0094] In multiple frames of environmental data, if there are tags associated with environmental data that are the same as the current state of the simulated vehicle, for example, the time associated with the environmental data is the same as the current time, and the location associated with the environmental data is the same as the current location; or the gear position of the vehicle associated with the environmental data is also the same as the current gear position of the simulated vehicle, then the environmental data is determined to be the current environmental data that matches the current state.

[0095] It should be noted that the test system in the embodiment of the present application is a closed-loop test system, which simulates a simulated vehicle. During the testing of the intelligent driving software, the simulated vehicle is controlled to perform intelligent driving actions. The specific execution results of the simulated vehicle are used as the test results of the intelligent driving software, which completely simulates the application scenarios of the real vehicle and realizes a closed-loop test.

[0096] During the test, the simulated vehicle simulates the driving of the real vehicle, and the simulated vehicle simulates the driving of the real vehicle, so in general, the driving time and position of the simulated vehicle and the real vehicle are the same. For example, the real vehicle drives to point A at time t1 to collect environmental data a, and the simulated vehicle also drives to point A at time t1 during the test based on the intelligent driving software. Then the time associated with environmental data a is time t1, and the current time of the simulated vehicle is time t1, so the time associated with the target environmental data collected at point A is the same as the current time. This embodiment obtains the current environmental data of the simulated vehicle from the environmental data collected by the real vehicle based on time and position synchronization, so that the test system can obtain environmental data that is closer to and more accurate than the real vehicle in the actual driving scene, so the intelligent driving software is tested based on accurate environmental data, and the test results are more accurate.

[0097] As an example, in some scenarios, it is difficult to achieve complete consistency between the control of the simulated vehicle and the control of the real vehicle, especially in some complex environments (such as parking scenarios), the difference between the simulated vehicle and the real vehicle may gradually increase (for example, during the parking process of the simulated vehicle, the parking path planned may deviate from the parking path of the real vehicle). For example, the real vehicle drives to point A at time t1 to collect environmental data a, while the simulated vehicle drives to point A at time t2 during the test of the intelligent driving software. Then the time associated with the environmental data a is time t1, and the current time of the simulated vehicle is time t2, so the time associated with the target environmental data collected at point A is different from the current time; or, the real vehicle drives to point A at time t1 to collect environmental data a, but the simulated vehicle does not pass through point A during the driving process, and there is a deviation in the position. Therefore, during the test process, it may be difficult to find environmental data that is synchronized in time and position at certain locations. For example, the driving path of the simulated vehicle is the same as the driving path of the real vehicle, but the driving speed is different, resulting in a difference in time. In this scenario, based on the current state of the simulated vehicle and the labels associated with multiple frames of environmental data, the current environmental data matching the current state is obtained, which can be specifically:

[0098] From the positions associated with multiple frames of environmental data, search for environmental data that is the same as the current position of the simulated vehicle. That is, position synchronization is given priority, and based on the current position of the simulated vehicle, the environmental data collected by the real vehicle at the same position is searched to determine the current environmental data that matches the current state.

[0099] The above method is suitable for testing driving scenarios on the road. In this driving scenario, the environmental data collected at the same position is usually one frame, so based on the consideration of position synchronization, the current environmental data of the simulated vehicle in the current state can be accurately obtained.

[0100] In some examples, if the actual vehicle moves back and forth at the same location, multiple frames of environmental data are collected at the same location. The environmental data determined based on location synchronization is multiple frames. In this scenario, the current environmental data that matches the current state is determined by combining other tags. For example, from the environmental data determined based on location synchronization, the environmental data with the smallest time interval with the current time is found and determined as the current environmental data that matches the current state.

[0101] For example, the driving path of the simulated vehicle deviates from the driving path of the real vehicle, resulting in a position deviation. In this scenario, based on the current state of the simulated vehicle and the labels associated with multiple frames of environmental data, the current environmental data matching the current state is obtained, which can be specifically:

[0102] From the positions associated with the multi-frame environmental data, find the environmental data with the smallest distance to the current position of the simulated vehicle, and then combine the current time to determine the current environmental data that matches the current state. The specific implementation method is as follows Figure 3 Embodiment and development description.

[0103] Optionally, at least two tags are not limited to time and location, but may also be other tags, such as the gear position of the vehicle, the vehicle speed, etc. Correspondingly, the current state of the simulated vehicle is not limited to the current location and the current time, but also includes the gear position and speed of the simulated vehicle, etc. Based on multiple tags, the matched current environment data is closer to the actual scene, and the test results of the intelligent driving software are more accurate.

[0104] Taking the gear position of the vehicle as an example, based on the current state of the simulated vehicle and the labels associated with the multi-frame environmental data, the current environmental data matching the current state is obtained, which can be specifically:

[0105] Based on the current gear position of the simulated vehicle, a plurality of frames of environmental data identical to the current gear position are obtained;

[0106] Based on the current time and current position of the simulated vehicle, and the collected time and position corresponding to the multi-frame environmental data, current environmental data matching the current state is obtained.

[0107] In this example, the current gear of the vehicle is used to filter out the environmental data that does not match the gear. Taking the parking scene as an example, during the parking process, the gears of the vehicle include reverse gear and forward gear, etc. If the current gear of the simulated vehicle is reverse gear, the environmental data associated with the forward gear can be filtered first, and only the environmental data associated with the reverse gear can be included. In this way, the amount of matching data can be reduced, and at the same time, matching based on the environmental data of the same location passed during the reverse and forward processes can be avoided.

[0108] S202, obtaining reconstructed environment data based on the reconstruction of the current environment data;

[0109] In this step, the scene is reconstructed based on the current environmental data collected by the real vehicle, realizing the conversion from the real vehicle driving scene to the simulated vehicle driving scene.

[0110] S203, controlling the simulated vehicle according to the reconstructed environment data to update the current state of the simulated vehicle.

[0111] In this step, after determining the reconstructed environment data, the intelligent driving software performs intelligent control on the simulated vehicle based on the reconstructed environment data to implement the test of the intelligent driving software. The control result of the simulated vehicle reflects the test result of the intelligent driving software under the reconstructed environment data.

[0112] In this embodiment, by formulating multiple labels for the recycled road sampling environmental data, when testing the intelligent driving software, the road sampling environmental data that is closer to the real-time state of the simulated vehicle can be matched based on the multiple labels associated with the environmental data based on the current state of the simulated vehicle. The driving scene of the simulated vehicle can be reconstructed based on the road sampling environmental data, and the intelligent driving software can plan and control the path according to the accurate driving scene to control the driving of the simulated vehicle. Since the driving scene has a high degree of matching with the current state of the simulated vehicle, the control result of the simulated vehicle can accurately reflect the test results reflected by the decision of the intelligent driving software in the corresponding driving scene, and the test results of the intelligent driving software are highly accurate.

[0113] Figure 3 This is a flow chart of a smart driving software testing method provided by another embodiment of the present application. This embodiment is based on the above embodiment and further proposes an implementation method of S201 on the basis of the above embodiment, such as Figure 3 As shown, S201 includes:

[0114] S301, searching for a target position with the shortest distance from the current position of the simulated vehicle from positions associated with multiple frames of environmental data;

[0115] In this step, if there is no position associated with the multiple frames of environmental data that is the same as the current position of the simulated vehicle (there is a deviation between the driving path of the simulated vehicle and the driving path of the real vehicle for collecting environmental data), the position associated with the environmental data that is the shortest distance from the current position is used as the target position.

[0116] If there is a position among the positions associated with the multi-frame environmental data that is the same as the current position of the simulated vehicle, the current position is used as the target position.

[0117] That is, the distance between the target position and the current position of the simulated vehicle is greater than or equal to 0.

[0118] S302, searching for a target time with a minimum time interval with the current time from the environmental data associated with the target location;

[0119] In this embodiment, after finding the environmental data corresponding to the target position, the current environmental data of the simulated vehicle in the current state is matched based on the time associated with the environmental data corresponding to the target position and the current time of the simulated vehicle.

[0120] As an example, the time interval between the time corresponding to each frame of environmental data associated with the target position and the current time is obtained; and the target time is determined based on the minimum time interval.

[0121] As an example, searching the target time with the smallest time interval with the current time from the environment data associated with the target location also includes:

[0122] From the environment data associated with the target location, find the target time that is after the current time and has the smallest time interval with the current time.

[0123] S303, determining current environment data matching the current state based on the target time;

[0124] In this step, the environmental data associated with the target time is determined as the current environmental data matching the current state.

[0125] The specific implementation process of this embodiment is described below with examples and drawings:

[0126] Take automatic parking as an example. Figure 4 As shown, assuming that the actual vehicle is in the process of automatic parking, the parking path is the path shown by the arrow in the figure. Among them, the actual vehicle passes through point B twice, so the actual vehicle collects the environmental data of point B at different times, and obtains at least two frames of environmental data. The environmental data of point B collected at time t1 is defined as the first environmental data, and the environmental data of point B collected at time t2 is defined as the second environmental data.

[0127] During the process of simulating the parking of a real vehicle by a simulated vehicle, if the target position is determined to be point B based on the current position of the simulated vehicle, then the associated environmental data including the first environmental data and the second environmental data are found based on the target position point B. If the current time corresponding to the current position of the simulated vehicle is t-1, the time t1 associated with the first environmental data and the time t associated with the second environmental data are both after t-1, and the time t is closer to the time t-1, so the time t1 is determined to be the target time, and then the first environmental data collected at the time t1 is determined to be the current environmental data matching the current state.

[0128] Through the above method, the driving error between the simulated vehicle and the real vehicle can be corrected to obtain accurate environmental data, and then the environmental data can be reconstructed, so that the intelligent driving software can control the simulated vehicle based on more accurate environmental data, thereby improving the accuracy of the test.

[0129] In this embodiment, the test of the intelligent driving software is more flexible. During the test, the driving path of the simulated vehicle can be adjusted according to the environmental data (a more reasonable driving path can be planned), and it does not need to be completely consistent with the actual driving path of the real vehicle. And when the actual driving paths of the simulated vehicle and the real vehicle are inconsistent, it is also possible to find the target position with the smallest distance from the current position of the simulated vehicle from the positions associated with the multi-frame environmental data, and find the target time with the smallest time interval with the current time from the environmental data associated with the target position, and then obtain the current environmental data closest to or most matching the current environment of the simulated vehicle based on the environmental data associated with the target time, so as to ensure the accuracy of the reconstructed scene data obtained by the intelligent driving software, and then ensure the accuracy of the test results.

[0130] Figure 5 This is a flow chart of a smart driving software testing method provided by another embodiment of the present application. This embodiment is based on all the above embodiments, and further proposes an implementation method of S202 on the basis of all the above embodiments, such as Figure 5 As shown, S202 includes:

[0131] S501, obtaining a static target based on current environment data;

[0132] It should be noted that when the real vehicle collects data on the road, each frame of environmental data collected includes static targets, dynamic targets and radar detection data. Therefore, based on the current environmental data, the static target corresponding to the current position can be obtained.

[0133] It is understood that static targets refer to absolutely stationary targets, such as lane lines, parking spaces, sensing detection relative position points (such as freespace-drivable area or accessible space), static obstacles, etc. Dynamic targets refer to target-level obstacles with an absolute speed not equal to 0, including walking people and moving cars.

[0134] S502, determining a dynamic target based on current environmental data obtained by the simulated vehicle in various states;

[0135] Since dynamic targets are changing, they cannot be identified and extracted through one frame of environmental data. Therefore, this step determines the dynamic target based on the current environmental data obtained by the simulated vehicle in various states. For example, assuming that the current time is t3, the dynamic target is determined based on the current environmental data obtained at t1 to t3, and the various states are the current state of the simulated vehicle at t1, the current state at t2, and the current state at t3.

[0136] Specifically, the dynamic target is extracted based on the time sequence of the environmental data.

[0137] S503, based on the simulated radar position and the simulated radar parameters, simulating and calculating radar detection data obtained when the simulated radar detects static targets and dynamic targets in the current state;

[0138] In this step, the acquired static targets and dynamic targets are used to simulate the targets around the simulated vehicle, and then based on the simulated radar position and simulated radar parameters on the simulated vehicle, the radar detection data obtained when the simulated radar detects the surrounding static targets and dynamic targets at the current position is simulated and calculated, and the radar detection data is used as the data detected by the simulated vehicle in the actual scene.

[0139] S504, obtaining reconstructed environment data based on the reconstruction of the static target, the dynamic target and the radar detection data.

[0140] In this step, based on the static targets, dynamic targets and radar detection data obtained above, a vehicle driving scene is constructed to obtain reconstructed environment data.

[0141] It should be noted that although dynamic targets, static targets and radar detection data can be obtained based on the environmental data collected by the real vehicle, there may be certain deviations between the real vehicle driving process and the simulation process of the simulated vehicle controlled by the intelligent driving software. If the dynamic targets and static targets in the environmental data are directly used, the deviation between the real vehicle and the simulated vehicle is ignored, affecting the accuracy of the test results of the intelligent driving software. Moreover, even if the radar position and radar parameters of the real vehicle are completely consistent with the simulated radar position and simulated radar parameters of the simulated vehicle, there will be deviations in the radar detection data based on the deviation between the real vehicle and the simulated vehicle. Therefore, using the environmental data collected by the real vehicle as the environmental data of the simulated vehicle will result in a large deviation in the test results of the intelligent software.

[0142] In this embodiment, when the current environment data matching the current state of the simulated vehicle is obtained, the static target is extracted. At the same time, the dynamic target is extracted based on the multi-frame current environment data. The static target and the dynamic target are corrected based on the deviation between the real vehicle and the simulated vehicle (the specific correction method can be referred to as follows Figure 6Embodiment), then based on the simulated radar position and the simulated radar parameters, the simulated radar detects these static targets and dynamic targets in the current state, and obtains radar detection data. On the one hand, the radar detection data is calculated based on the corrected static targets and dynamic targets, and the radar data deviation caused by the deviation between the real vehicle and the simulated vehicle is corrected; on the other hand, based on the simulated radar position and the simulated radar parameters, the radar detection data obtained by detecting these static targets and dynamic targets is simulated and calculated. It does not require that the simulated radar position and the simulated radar parameters must be completely consistent with the radar position and radar parameters on the real vehicle, and can be applied to different radars, and equipment deviation is eliminated.

[0143] Figure 6 This is a flow chart of a smart driving software testing method provided by another embodiment of the present application. This embodiment is based on all the above embodiments, and further proposes another implementation method of S202 on the basis of all the above embodiments, such as Figure 6 As shown, S202 includes:

[0144] S601, correcting current environment data based on the body coordinate system of the simulated vehicle;

[0145] When real vehicle road data is collected, the output environmental data is based on the body coordinate system of the real vehicle, that is, the coordinates of the target in the environmental data are relative to the body coordinate system of the real vehicle. For example, there is a pillar within 0.5m in front of the right side of the real vehicle, and the position of this pillar is relative to the body coordinate system of the real vehicle.

[0146] Since the simulated path of the simulated vehicle may deviate from the real vehicle during the simulation process, the position of the pillar in the current environmental data obtained is the coordinate relative to the body coordinate system of the real vehicle. If the environmental data of the simulated vehicle is directly constructed with this coordinate, the relative position of the pillar and the simulated vehicle is inaccurate. Therefore, the current environmental data needs to be corrected to obtain environmental data in the body coordinate system that conforms to the simulated vehicle.

[0147] As an example, the current environment data can be modified in the following ways:

[0148] According to the driving information of the actual vehicle when the current environmental data is collected, a first conversion matrix between the body coordinates of the actual vehicle and the preset coordinates is determined; based on the first conversion matrix, the current environmental data is converted to the preset coordinates.

[0149] Optionally, the preset coordinate system is an absolute coordinate system. The first conversion matrix is ​​different based on the driving information of the actual vehicle. The driving information includes the vehicle position, vehicle heading angle and speed. For static targets in the current environmental data, the speed is 0, and the first conversion matrix has a conversion value of 0 for the speed; for static targets without heading angle attributes, such as lane line points, parking spots, ultrasonic clustering points, freespace, etc., the vehicle heading angle is 0, that is, in the first conversion matrix, the conversion values ​​of the heading angle and speed are 0.

[0150] According to the current driving information of the simulated vehicle, a second conversion matrix between the body coordinates of the simulated vehicle and the preset coordinates is determined; based on the second conversion matrix, the environmental data under the preset coordinate system is converted to the body coordinate system of the simulated vehicle to obtain the corrected current environmental data.

[0151] Likewise, the second conversion matrix varies based on different driving information of the simulated vehicle.

[0152] In this embodiment, the current environment data is corrected by the above correction method, so that the data deviation between the real vehicle and the simulated vehicle can be corrected, so that the corrected current environment data is in the body coordinates of the simulated vehicle with high accuracy.

[0153] As an example, the current environment data under the body coordinates of the simulated vehicle can be obtained in the following way:

[0154]

[0155] Among them, objx new objx is the lateral coordinate of the target in the current environment data in the body coordinate system of the simulated vehicle after correction; new objx is the corrected longitudinal coordinate of the target in the current environment data in the body coordinate system of the simulated vehicle; new The heading angle of the target in the current environment data in the body coordinate system of the simulated vehicle after correction.

[0156] objx ori The lateral coordinates of the target in the current environment data before correction in the body coordinates of the real vehicle; objy ori objθ is the longitudinal coordinate of the target in the current environment data before correction in the body coordinates of the real vehicle; ori Correct the heading angle of the target in the current environment data in the body coordinates of the real vehicle;

[0157] vehx ori is the lateral coordinate of the body coordinate of the real vehicle in the preset coordinate system; ori is the longitudinal coordinate of the body coordinate of the real vehicle in the preset coordinate; vehθ orivehy ori is the heading angle of the body coordinates of the actual vehicle in the preset coordinates;

[0158] vehx new It is the lateral coordinate of the body coordinate of the simulated vehicle in the absolute coordinate system; new is the longitudinal coordinate of the simulated vehicle body in the absolute coordinate system; vehθ new It is the heading angle of the longitudinal coordinate of the body coordinate of the simulated vehicle in the absolute coordinate system.

[0159] S602, reconstructing based on the corrected current environment data to obtain reconstructed environment data.

[0160] In this step, the corrected current environment data is reconstructed. The specific reconstruction process is the same as above. Figure 5 The embodiments are the same and will not be described again here.

[0161] In this step, before reconstructing the environmental data of the simulated vehicle, the current environmental data is first converted to the body coordinates of the simulated vehicle to adapt to the position, heading, etc. of the simulated vehicle and the real vehicle. When there is a deviation, the simulated vehicle can obtain accurate environmental data under its own coordinates, which can avoid misjudgment or error of the intelligent driving software and improve the accuracy of the intelligent driving software test results.

[0162] Figure 7 This is a flow chart of a smart driving software testing method provided by another embodiment of the present application. This embodiment is based on all the above embodiments, and further proposes an implementation method of S203 on the basis of all the above embodiments, such as Figure 7 As shown, S203 includes:

[0163] S701, obtaining a simulated vehicle control instruction according to the reconstructed environment data;

[0164] In this step, the intelligent driving software intelligently controls the simulated vehicle according to the reconstructed environment data, and obtains the test result of the intelligent driving software according to the control status of the simulated vehicle. For example, in the reconstructed environment data, there is an obstacle in front of the simulated vehicle, and the test result of the intelligent driving software is judged based on the intelligent driving software controlling the simulated vehicle to brake, decelerate or avoid obstacles.

[0165] In a further embodiment, after the information to be driven of the simulated vehicle is obtained based on the reconstructed environment data, the control instructions corresponding to the information to be driven of the simulated vehicle are calculated based on the perfect control algorithm.

[0166] Optionally, the specific implementation method may be:

[0167] The to-be-traveled information of the simulated vehicle is updated according to the reconstructed environment data, and the to-be-traveled information includes a trajectory point set, vehicle speed, acceleration, gear position, etc.

[0168] Optionally, the trajectory point set is pre-planned based on the initial position and target position of the simulated vehicle. During the driving process of the simulated vehicle, the unpassed trajectory points will be updated based on the reconstructed environment data in the current state, so that the simulated vehicle can reach the target position accurately and quickly without collision. In addition, during the driving process of the simulated vehicle, the vehicle speed, acceleration or gear position will be updated according to the reconstructed environment data, for example, braking, deceleration, acceleration or shifting.

[0169] The perfect control algorithm is a perfect follow-up control based on the trajectory formed by the intelligent driving software based on the trajectory point set. The process of perfect follow-up control is as follows:

[0170] First, according to the vehicle speed and acceleration of the simulated vehicle, the travel distance of the simulated vehicle after a preset time step is determined; according to the current trajectory point where the current position of the simulated vehicle is located and the travel distance, the next target trajectory point is determined; according to the next target trajectory point and the current trajectory point, the lateral change, longitudinal change and heading angle change of the simulated vehicle are determined; according to the lateral change, longitudinal change and heading angle change, the target lateral coordinate, target longitudinal coordinate and target heading angle of the simulated vehicle are obtained; according to the target lateral coordinate, target longitudinal coordinate and target heading angle, a control instruction of the simulated vehicle is generated, and the control instruction is used to control the heading and speed of the simulated vehicle.

[0171] Optionally, the preset time step can be determined according to a preset test frame rate, that is, after each time step, the current state of the simulated vehicle is updated, the current environmental data of the simulated vehicle is reacquired, the scene is reconstructed, and the intelligent driving software is tested.

[0172] like Figure 8 As shown, assuming that the current trajectory points of the simulated vehicle are trajectory points 3 and 4, the calculated travel distance of the simulated vehicle is X (assuming that X is less than the distance between the two trajectory points), and the next target trajectory point is determined to be trajectory point 6. Then, based on trajectory points 3, 4 and 6, the corresponding coordinates and heading angles when the simulated vehicle travels to trajectory point 6 are calculated, and the control instructions of the simulated vehicle are generated according to the calculated results to control the simulated vehicle to accurately travel to the target trajectory point.

[0173] The following is a specific algorithm to illustrate the perfect control of the simulated vehicle in the embodiment of the present application:

[0174] The trajectory pre-planned based on the initial position and target position of the simulated vehicle is a trajectory in absolute coordinates (global coordinates). The test system traverses the trajectory in advance and converts the trajectory to the body coordinate system (vcs coordinate system) of the simulated vehicle to obtain a discrete set of trajectory points. Figure 8 As shown, the center point of the rear axle of the simulated vehicle is taken as the origin, the center direction of the front of the vehicle is the positive direction of the x-axis, and the y-axis is rotated 90 / degree counterclockwise from the x-axis to obtain the vehicle body coordinate system.

[0175] Based on the vehicle body coordinate system, the coordinate transformation of the i-th trajectory point is as follows:

[0176]

[0177] in, They are respectively the longitudinal coordinate, lateral coordinate and heading angle of the center point of the rear axle of the simulated vehicle in the absolute coordinate system; are the longitudinal coordinate and transverse coordinate of the i-th trajectory point in the absolute coordinate system respectively; are respectively the longitudinal coordinate and the lateral coordinate of the i-th trajectory point in the body coordinates of the simulated vehicle.

[0178] Traverse through the following conditions:

[0179] If points vcs [i]<0;

[0180] FrontIndex = i + 1;

[0181] After the traversal is completed, FrontIndex is the index of the first track point in front of the current position of the simulated vehicle before it is updated (point 4 in the figure), which is recorded as points vcs [FrontIndex];

[0182] Optionally, the vehicle travel distance is calculated by the following formula:

[0183]

[0184] Among them, VehSpd is the real-time speed of the vehicle; Δt is the preset time step; LongAcc is the longitudinal acceleration of the vehicle; ΔDist is the distance traveled by the vehicle within the time step.

[0185] Optionally, the longitudinal change of the simulated vehicle and the change of the heading angle are calculated by the following method:

[0186] In order to make the vehicle change continuously, the current position of the simulated vehicle and the front trajectory are fitted with a third-order polynomial to obtain the coefficients [a, b, c, d];

[0187] When the curvature of the road where the simulated vehicle is located is ≤ 0.001, that is, |points[FrontIndex].curvature|<0.001:

[0188] Δx = ΔDist;

[0189] That is, when the curvature between the next target trajectory point and the current trajectory point is small, the road section is a straight line. In this case, the longitudinal change of the simulated vehicle is the travel distance of the simulated vehicle. If the curvature is large, the road section is a curve, and the simulated vehicle needs to be controlled to turn. Therefore, the simulated vehicle includes longitudinal change and lateral change. In this case, the longitudinal change and lateral change of the simulated vehicle are calculated in the following way:

[0190]

[0191] Calculate Δy and Δθ based on the results of the polynomial fitting;

[0192] Δy=a×Δx 3 +b×Δx 2 +c×Δx+d;

[0193] Δθ1=arctan(3a×Δx 2 +2b×Δx+c);

[0194] Δθ2=ΔDist×points[FrontIndex].curvature;

[0195] If: Δθ1×Δθ2≤0, then: Δθ=0;

[0196] If not, then:

[0197]

[0198] Among them, Δx is the longitudinal change of the simulated vehicle, Δy is the lateral change of the simulated vehicle, and Δθ is the heading angle change of the simulated vehicle.

[0199] In this example, Δθ1 and Δθ2 are the heading angle changes of the simulated vehicle calculated by two different methods. Then, the heading angle change Δθ of the simulated vehicle is calculated by the two methods to improve the calculation accuracy.

[0200] Optionally, based on the longitudinal change, lateral change and heading angle change of the simulated vehicle, the posture change relative to the current posture can be obtained.

[0201] Furthermore, the new posture of the simulated vehicle is calculated based on the posture change of the simulated vehicle:

[0202]

[0203] Among them, veh x ' is the updated x coordinate of the simulated vehicle (target longitudinal coordinate); veh y ' is the updated y coordinate of the simulated vehicle (target lateral coordinate); veh θ ′ is the updated heading angle of the simulated vehicle (target heading angle). x ',veh y ′ and veh θ ′ represents the new pose of the simulated vehicle.

[0204] S702, controlling the simulated vehicle according to the control instruction to update the current state of the simulated vehicle.

[0205] In this step, the intelligent driving software controls the simulated vehicle to travel to the next target trajectory point according to the control instructions. After reaching the next target trajectory point, the current state of the simulated vehicle is updated, and then based on the updated current state, the steps in the above embodiments are returned to continue testing the intelligent driving software.

[0206] In this embodiment, after obtaining the reconstructed environmental data, the simulated vehicle is controlled by considering the most perfect driving state of the vehicle, which can avoid the influence of some additional factors on the control, ensure that the control of the simulated vehicle is more matched with the decision response of the intelligent driving software, better reflect the test results of the intelligent driving software, and improve the reliability of the test results.

[0207] The present application also provides an intelligent driving software testing system, the testing system comprising:

[0208] A data matching module, for obtaining current environmental data matching the current state based on the current state of the simulated vehicle and tags associated with multiple frames of environmental data, wherein each frame of environmental data is associated with at least two tags, and the at least two tags respectively reflect the time and location of each frame of environmental data collection, and the current state includes the current location and the current time;

[0209] A scene construction module is used to reconstruct based on current environment data to obtain reconstructed environment data;

[0210] An intelligent driving module is used to determine the to-be-driven information of the simulated vehicle according to the reconstructed environment data and the current state of the simulated vehicle, the to-be-driven information including the trajectory point set, vehicle speed, acceleration and gear position, etc.;

[0211] The control module is used to update the current state of the simulated vehicle according to the to-be-traveled information.

[0212] Optionally, the control module is a closed-loop control module.

[0213] Furthermore, the test system also includes a data analysis module for analyzing the road sampling data, extracting and storing each frame of environmental data according to the proposed label, so as to facilitate scene reconstruction and extraction. Based on the analysis module, environmental data extraction and label association are realized.

[0214] Optionally, the intelligent driving module is a module composed of intelligent driving software, which is used to receive the reconstructed environment data output by the scene construction module, calculate the reconstructed environment data, and output the information to be driven, which includes the trajectory point set, vehicle speed, acceleration and gear, etc. The control module is used to receive the output result of the intelligent driving module, and then update the current state of the vehicle based on perfect control, such as controlling the vehicle to drive in a simulated environment; and collect the position, gear, speed, heading angle and other information of the vehicle during driving, and update the current state of the simulated vehicle in real time.

[0215] Optionally, the intelligent driving module can be activated by the following methods:

[0216] As in one example, the status information based on the road sampling data is activated, and this method is used to reproduce the road sampling data scenario in the simulation test.

[0217] As in another example, manual activation is used to generalize road data sampling scenarios in simulation tests.

[0218] Optionally, the test system also includes a loop scheduling module and a data recording module. The loop scheduling module is used to loop schedule the above modules according to an easily set frame rate to execute the test process. The data recording module is used to record and save new data generated during the test process to facilitate problem troubleshooting and automatic evaluation.

[0219] In the embodiment of the present application, the optimal environmental data that best suits the real-time state of the simulated vehicle is selected based on time-space synchronization, and the environmental data is corrected according to the real-time state of the simulated vehicle, and the corrected environmental data is sent to the parking algorithm module, and the output result of the parking algorithm is updated in a closed loop based on perfect control, and the optimal environmental data that best suits the real-time state of the simulated vehicle is selected again based on time-space synchronization according to the real-time state of the closed-loop simulated vehicle, and the cycle is repeated. A closed-loop test of the automatic parking algorithm's decision-making response to the road-collected environmental data is achieved.

[0220] The intelligent driving software testing system provided in the embodiment of the present application can execute the above-mentioned method embodiment. Its specific implementation principles and technical effects can be found in the above-mentioned method embodiment, and this embodiment will not be repeated here.

[0221] The embodiment of the present application further provides an electronic device 900, comprising: a processor 901, a memory 902; optionally, the electronic device 900 further comprises a communication component 903. The processor 901, the memory 902 and the communication component 903 are connected via a CAN bus.

[0222] The memory 902 is used to store computer programs / instructions; the processor 901 is used to execute the computer programs / instructions stored in the memory to implement the methods involved in the above embodiments.

[0223] The electronic device 900 also includes a communication interface, wherein the processor is used to provide computing power and control capabilities, and can be a GPU, CPU, NPU, MCU, FPGA, etc. The storage device includes an internal memory and a non-volatile memory. The non-volatile memory stores a computer program that implements the above method. The internal memory provides an environment for program startup and operation. The communication interface is used to communicate with an external terminal by wire or wireless.

[0224] In this embodiment, the electronic device may be a testing device for intelligent driving software.

[0225] The present invention also provides a computer-readable storage medium / computer program product, in which computer control instructions are stored / the computer program product includes computer control instructions, and when the computer control instructions are executed by a processor, they are used to implement the methods involved in the above-mentioned embodiments.

[0226] The above embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Any equivalent substitution or change made by a person skilled in the art based on the present invention is within the protection scope of the present invention.

Claims

1. A method for testing intelligent driving software, characterized in that: include: Based on the current state of the simulated vehicle and the tags associated with multiple frames of environmental data, current environmental data matching the current state is obtained, wherein each frame of the environmental data is associated with at least two tags, and the at least two tags respectively reflect the time and location of acquisition of each frame of the environmental data, and the current state includes the current location and the current time; Reconstructing based on the current environment data to obtain reconstructed environment data; The simulated vehicle is controlled according to the reconstructed environment data to update the current state of the simulated vehicle.

2. The method according to claim 1, characterized in that Based on the current state of the simulated vehicle and the labels associated with the multiple frames of environmental data, current environmental data matching the current state is obtained, including: Finding a target position with the shortest distance from the current position of the simulated vehicle from the positions associated with the multiple frames of environmental data; Based on the environmental data associated with the target location and the current time, current environmental data matching the current state is obtained.

3. The method according to claim 2, characterized in that Based on the environmental data associated with the target location and the current time, obtaining current environmental data matching the current state includes: Finding a target time having a minimum time interval with the current time from the environmental data associated with the target location; Based on the target time, current environment data matching the current state is determined.

4. The method according to any one of claims 1 to 3, characterized in that: The step of obtaining current environmental data matching the current state based on the current state of the simulated vehicle and the tags associated with the multiple frames of environmental data includes: Based on the current gear position of the simulated vehicle, multiple frames of environmental data identical to the current gear position are obtained, wherein the current state also includes the current gear position, and the at least two tags also reflect the vehicle gear position when each frame of the environmental data is collected; Based on the current time and current position of the simulated vehicle, and the collected time and position corresponding to the multiple frames of environmental data, current environmental data matching the current state is obtained.

5. The method according to any one of claims 1 to 3, characterized in that: The environmental data includes static targets and dynamic targets. The reconstructing based on the current environmental data to obtain the reconstructed environmental data includes: Based on the current environment data, a static target is obtained; Determining a dynamic target based on current environmental data obtained by the simulated vehicle in various states; Based on the simulated radar position and the simulated radar parameters, simulating and calculating the radar detection data obtained when the simulated radar detects the static target and the dynamic target in the current state; Based on the reconstruction of the static target, the dynamic target and the radar detection data, reconstructed environment data is obtained.

6. The method according to any one of claims 1 to 3, characterized in that: The step of reconstructing based on the current state of the simulated vehicle to obtain reconstructed environmental data comprises: Based on the body coordinate system of the simulated vehicle, correcting the current environment data; Reconstruction is performed based on the corrected current environment data to obtain reconstructed environment data.

7. The method according to claim 6, characterized in that The correcting the current environment data based on the body coordinates of the simulated vehicle includes: Determine a first conversion matrix between the body coordinates of the actual vehicle and the preset coordinates according to the driving information of the actual vehicle when the current environmental data is collected; Based on the first conversion matrix, converting the current environment data to the preset coordinates; Determine a second conversion matrix between the body coordinates of the simulated vehicle and the preset coordinates according to the current driving information of the simulated vehicle; Based on the second conversion matrix, the environmental data in the preset coordinate system is converted to the body coordinate system of the simulated vehicle to obtain corrected current environmental data; The driving information includes at least one of a vehicle position, a vehicle heading angle and a vehicle speed.

8. The method according to any one of claims 1 to 3, characterized in that: Before obtaining the current environmental data matching the current state based on the current state of the simulated vehicle and the tags associated with the multi-frame environmental data, the method further includes: Initial road data collection is obtained, and the collection information of the initial road data is used as a label, and the label and the initial road data are associated. The collection information of each frame of environmental data includes the time and location of collection.

9. The method according to any one of claims 1 to 3, characterized in that: The controlling the simulated vehicle according to the reconstructed environment data to update the current state of the simulated vehicle comprises: Obtaining the simulated vehicle control instruction according to the reconstructed environment data; According to the control instruction, the simulated vehicle is controlled to update the current state of the simulated vehicle.

10. The method according to claim 9, characterized in that The step of obtaining the simulated vehicle control instruction according to the reconstructed environment data comprises: updating the to-be-traveled information of the simulated vehicle according to the reconstructed environment data, wherein the to-be-traveled information includes one or more of a trajectory point set, a vehicle speed, an acceleration, and a gear position; Determining a travel distance of the simulated vehicle after a preset time step according to the vehicle speed and the acceleration; Determine a next target trajectory point according to the current trajectory point where the current position of the simulated vehicle is located and the travel distance; Determining a lateral change, a longitudinal change, and a heading angle change of the simulated vehicle according to the next target trajectory point and the current trajectory point; A control instruction of the simulated vehicle is obtained according to the lateral variation, the longitudinal variation and the heading angle variation, and the control instruction is used to control the heading and speed of the simulated vehicle.

11. An intelligent driving software testing system, characterized in that: The test system comprises: A data matching module, for obtaining current environmental data matching the current state based on the current state of the simulated vehicle and tags associated with multiple frames of environmental data, wherein each frame of the environmental data is associated with at least two tags, and the at least two tags respectively reflect the time and location of the collection of each frame of the environmental data, and the current state includes a current location and a current time; A scene construction module, used for reconstructing based on the current environment data to obtain reconstructed environment data; An intelligent driving module, used for determining the to-be-driven information of the simulated vehicle according to the reconstructed environment data and the current state; The control module is used to update the current state of the simulated vehicle according to the to-be-traveled information.

12. An electronic device, characterized in that: include: Memory, processor; The memory is used to store computer programs / instructions; The processor is configured to implement the method according to any one of claims 1 to 10 according to the computer program / instructions stored in the memory.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program / instruction, and the computer program / instruction is used to implement the method according to any one of claims 1 to 10 when executed by a processor.

14. A computer program product, characterized in that The computer program product comprises a computer program / instructions, which are used to implement the method according to any one of claims 1 to 10 when executed by a processor.

Images