Vehicle parking test case generation method, parking test method and computer equipment

Abstract

The invention provides a vehicle parking test case generation method, a parking test method and computer equipment, and the vehicle parking test case generation method comprises the steps: determining a target motion track of a vehicle, and according to a plurality of environment elements sensed by the vehicle in different parking scenes and the element information of each environment element, determining the target motion track of the vehicle; generating a plurality of groups of target environment elements corresponding to each frame of motion trail and element information of each group of target environment elements, generating a plurality of target parking test scenes according to the plurality of groups of target environment elements corresponding to each frame of motion trail and the element information of each group of target environment elements, and performing parking test according to the plurality of target parking test scenes. And generating a plurality of parking test cases for performing a vehicle parking test. Therefore, based on one motion track, multiple parking test scenes can be generated, and the complexity and diversity of the parking test scenes are improved, so that the complexity and diversity of the parking test cases are improved, the vehicle parking test is more comprehensive, and the vehicle parking function test is more complete.

Description

Technical Field

[0001] This application relates to the field of vehicle parking testing technology, and in particular to a vehicle parking test case generation, parking testing method, and computer equipment. Background Technology

[0002] With the deep integration of artificial intelligence, the Internet of Things, chip technology and the automotive industry, vehicle intelligence has become the core direction for the transformation and upgrading of the global automotive industry.

[0003] Assisted parking is a crucial function in automotive intelligence. The parking algorithms involved in assisted parking systems are divided into perception algorithms and planning and control algorithms. The planning and control algorithms' functions, such as trajectory planning and replanning, rely on environmental elements output by the perception algorithms, including parking spaces, static obstacles, curbs, and dynamic obstacles. In the early stages of project development, the perception algorithms are not yet mature, and their performance requires long-term iteration to stabilize. Directly relying on environmental elements perceived by the perception algorithms in real-world driving scenarios to validate the planning and control algorithms would require multiple real-world driving perceptions after the perception algorithm technology matures, thus significantly slowing down the development and iteration pace of the parking planning and control algorithms. Therefore, it is urgent to encapsulate parking simulation scenarios into parking test cases and input these test cases into the planning and control algorithms before the perception algorithms mature, in order to verify the performance of the planning and control algorithms in advance.

[0004] In existing technologies, parking simulation scenarios are generated by simply arranging and combining environmental elements such as parking spaces, obstacles, and walls collected from real vehicles in parking scenarios to generate static parking scenarios. This method produces simple and static parking simulation scenarios, which can only verify the initial path planning capability of the planning and control algorithm in parking scenarios. It cannot verify the algorithm's ability to dynamically plan and replan its trajectory when dynamic obstacles are present or when obstacles change abruptly during vehicle movement.

[0005] Therefore, existing vehicle parking testing methods suffer from the problem that the generated parking simulation scenarios are simple and static, resulting in simple and incomplete parking test cases, which in turn leads to incomplete vehicle parking testing and incomplete vehicle parking function testing. Summary of the Invention

[0006] This application provides a vehicle parking test case generation, parking test method, and computer device. Based on multiple environmental elements perceived in different parking scenarios and the element information of each environmental element, it generates multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to each frame of the vehicle's target motion trajectory. This generates multiple target parking test scenarios, thus enabling the generation of multiple parking test scenarios corresponding to different target motion trajectories of the vehicle. This makes the parking test scenarios complex and diverse, resulting in complex and diverse parking test cases. Consequently, the vehicle parking test is more comprehensive, capable of testing multiple functions of vehicle parking as much as possible, and making the vehicle parking function test more complete.

[0007] In a first aspect, embodiments of this application provide a method for generating vehicle parking test cases. The method includes: determining multiple environmental elements perceived by the vehicle in different parking scenarios and the element information of each environmental element; determining the vehicle's target motion trajectory; generating multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to each frame of the target motion trajectory based on the multiple environmental elements and the element information of each environmental element; generating multiple target parking test scenarios based on the multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to each frame of the motion trajectory; and generating multiple parking test cases based on the multiple target parking test scenarios for use in vehicle parking testing.

[0008] By employing the aforementioned technical solution, multiple environmental elements perceived by the vehicle in different parking scenarios, along with their element information, are determined. Based on these environmental elements and their element information, multiple sets of target environmental elements and their element information are generated corresponding to each frame of the vehicle's target motion trajectory. Multiple target parking test scenarios are then generated based on these multiple sets of target environmental elements and their element information for each frame of the motion trajectory. This allows for the generation of multiple parking test scenarios where environmental elements change along the vehicle's motion trajectory. Furthermore, multiple parking test cases are generated from these multiple target parking test scenarios for vehicle parking testing. Therefore, multiple parking test scenarios can be generated along a single motion trajectory, increasing the complexity and diversity of the parking test scenarios, thereby increasing the complexity and diversity of the generated parking test cases. This makes vehicle parking testing more comprehensive, enabling the testing of multiple vehicle parking functions as much as possible, resulting in more complete vehicle parking function testing.

[0009] In one possible implementation of the first aspect above, generating multiple sets of target environment elements and element information of each set of target environment elements corresponding to each frame of motion trajectory along the target motion trajectory, based on multiple environment elements and element information of each environment element, and trajectory attributes of the first frame of motion trajectory in the target motion trajectory, includes: generating multiple sets of target environment elements and element information of each set of target environment elements corresponding to the first frame of motion trajectory based on multiple environment elements and element information of each environment element, and trajectory attributes of the (k-1)th frame of motion trajectory in the target motion trajectory, and trajectory attributes of the kth frame of motion trajectory, generating multiple sets of target environment elements and element information of each set of target environment elements corresponding to the kth frame of motion trajectory in the target motion trajectory, where k ranges from 2 to T, and T is the total number of frames of motion trajectory included in the target motion trajectory.

[0010] By adopting the above technical solution, based on multiple environmental elements and their element information, as well as the trajectory attributes of the first frame of the target motion trajectory, multiple sets of target environmental elements and their element information are randomly generated corresponding to the first frame of the target motion trajectory. Based on these multiple environmental elements and their element information, the multiple sets of target environmental elements corresponding to the (k-1)th frame of the target motion trajectory, and their element information, as well as the trajectory attributes of the kth frame of the target motion trajectory, multiple sets of target environmental elements and their element information corresponding to the kth frame of the target motion trajectory are generated. This establishes a correlation between the multiple sets of target environmental elements corresponding to the kth frame of the motion trajectory and the multiple sets of target environmental elements corresponding to the (k-1)th frame of the motion trajectory, improving the rationality of the parking test scenario. Furthermore, considering the trajectory attributes of the vehicle's motion trajectory, the target environmental elements in the parking test scenario better match the perceived posture of the vehicle's actual parking trajectory, further enhancing the realism of the vehicle parking test scenario.

[0011] In one possible implementation of the first aspect above, the environment elements include static environment elements and dynamic environment elements. Based on multiple environment elements and their element information, multiple sets of target environment elements corresponding to the (k-1)th frame motion trajectory in the target motion trajectory and their element information, and the trajectory attributes of the k-th frame motion trajectory in the target motion trajectory, the generation of multiple sets of target environment elements and their element information corresponding to the k-th frame motion trajectory includes: updating the dynamic environment elements and their element information included in the multiple sets of target environment elements corresponding to the (k-1)th frame motion trajectory, based on the dynamic environment elements included in the multiple sets of target environment elements corresponding to the (k-1)th frame motion trajectory and their element information, and the trajectory attributes of the k-th frame motion trajectory, to generate multiple sets of target dynamic environment elements and their element information corresponding to the k-th frame motion trajectory, and / or based on multiple environment element packages... Based on the dynamic environment elements and element information of each dynamic environment element, and the trajectory attributes of the motion trajectory in the k-th frame, multiple sets of target dynamic environment elements and element information of each set of target dynamic environment elements corresponding to the motion trajectory in the k-th frame are generated. Based on the static environment elements included in the multiple environment elements and element information of each static environment element, the multiple sets of target static environment elements included in the motion trajectory in the (k-1)-th frame and element information of each static environment element, and the trajectory attributes of the motion trajectory in the k-th frame, multiple sets of target static environment elements and element information of each set of target static environment elements corresponding to the motion trajectory in the k-th frame are generated. Based on the multiple sets of target dynamic environment elements and element information of each set of target dynamic environment elements corresponding to the motion trajectory in the k-th frame, and the multiple sets of target static environment elements and element information of each set of target static environment elements corresponding to the motion trajectory in the k-th frame, multiple sets of target environment elements and element information of each set of target static environment elements corresponding to the motion trajectory in the k-th frame are obtained.

[0012] By adopting the above technical solution, the multiple sets of target dynamic environment elements corresponding to the motion trajectory of the k-th frame are related to the dynamic environment elements included in the multiple sets of target environment elements corresponding to the motion trajectory of the (k-1)-th frame, or are new dynamic environment elements among multiple environment elements. The multiple sets of target static environment elements corresponding to the motion trajectory of the k-th frame are related to the static environment elements included in the multiple sets of target environment elements corresponding to the motion trajectory of the (k-1)-th frame, or are new static environment elements among multiple environment elements. In this way, the environmental elements corresponding to the motion trajectory of each frame in the parking test scenario become complex and varied, thereby increasing the complexity and diversity of the parking test scenario.

[0013] In one possible implementation of the first aspect above, multiple target parking test scenarios are generated based on multiple sets of target environment elements corresponding to each frame of motion trajectory and the element information of each set of target environment elements. This includes: determining multiple sets of target environment elements within the field of view of the target camera corresponding to the vehicle from among the multiple sets of target environment elements corresponding to each frame of motion trajectory, thereby obtaining multiple sets of visible target environment elements corresponding to each frame of motion trajectory; performing perception jump simulation on the element information of the multiple sets of visible target environment elements corresponding to each frame of motion trajectory to generate target element information of the multiple sets of visible target environment elements corresponding to each frame of motion trajectory; and generating multiple target parking test scenarios based on the multiple sets of visible target environment elements corresponding to each frame of motion trajectory and the target element information corresponding to each set of visible target environment elements.

[0014] Using the above technical solution, multiple sets of visible target environment elements within the field of view of the target camera corresponding to the vehicle are extracted from multiple sets of target environment elements corresponding to each frame of motion trajectory. Perceptual jump simulation is then performed on the element information of these multiple sets of visible target environment elements. Based on the multiple sets of visible target environment elements corresponding to each frame of motion trajectory and the target element information corresponding to each set of visible target environment elements, multiple target parking test scenarios are generated. In this way, by simulating the camera field of view and perceptual jump, the parking test scenarios more closely resemble the real perception effect of the vehicle perception algorithm, improving the realism of the parking test scenarios and enabling more accurate vehicle parking tests.

[0015] In one possible implementation of the first aspect above, the method further includes generating a target parking test scenario in the following manner: determining scenario generation requirements; inputting the scenario generation requirements, multiple environmental elements and element information of each environmental element, and the vehicle's motion trajectory into a scenario generation model, so that the scenario generation model determines the vehicle's target motion trajectory; generating multiple sets of target environmental elements and element information of each set of target environmental elements corresponding to each frame of the target motion trajectory based on the multiple environmental elements and element information of each set of target environmental elements; and generating multiple target parking test scenarios corresponding to the scenario generation requirements based on the multiple sets of target environmental elements and element information of each set of target environmental elements corresponding to each frame of the motion trajectory.

[0016] By adopting the above technical solution, based on the scene generation model, multiple target parking scenes corresponding to the scene generation requirements are generated according to the scene generation requirements, multiple environmental elements, element information of each environmental element, and the vehicle's movement trajectory. This reduces the threshold for scene construction and improves the efficiency of parking test scene generation.

[0017] In one possible implementation of the first aspect above, determining multiple environmental elements perceived by the vehicle in different parking scenarios and the element information of each environmental element includes: determining multiple driving image data of the vehicle under different driving attributes in different parking scenarios, obtaining multiple driving image data, the driving attributes including driving position, driving angle and driving pose; performing perception processing on the multiple driving image data based on a perception algorithm to obtain multiple environmental elements of the vehicle in different parking scenarios and the element information of each environmental element, the element information of each environmental element including multiple element information corresponding to different driving attributes.

[0018] In one possible implementation of the first aspect mentioned above, the element information of the environmental elements includes the attribute information, location information, and relative position information of each environmental element and the vehicle.

[0019] By adopting the above technical solution, the system repeatedly collects data from different positions, angles, and relative poses in different parking scenarios, thereby restoring the real characteristics of the perception results as distance and angle change, covering multiple types, forms, and colors of similar environmental elements, and improving the diversity of scene environmental elements.

[0020] Secondly, this application also discloses a vehicle parking test method, which involves determining multiple parking test cases, generating the multiple parking test cases based on the vehicle parking test case generation method provided by any of the implementation methods in the first aspect, and performing vehicle parking tests based on each parking test case.

[0021] In one possible implementation of the first aspect above, vehicle parking testing is performed based on each parking test case, including: encapsulating each parking test case into a corresponding test container, the test container including a rule control algorithm; at least one test container injects the corresponding parking test case into the corresponding rule control algorithm so that the rule control algorithm simulates vehicle parking to achieve vehicle parking testing.

[0022] By adopting the above technical solution, a multi-container cluster is started to perform parking simulation tests in batches, thereby improving the efficiency of parking tests.

[0023] Thirdly, this application also discloses a computer device, which includes: a processor and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to enable the computer device to implement the vehicle parking test case generation method disclosed in any of the implementations of the first aspect above, and / or implement the vehicle parking test method disclosed in any of the implementations of the second aspect above.

[0024] The relevant beneficial effects of the third aspect mentioned above can be found in the relevant descriptions of the first and second aspects mentioned above, and will not be repeated here. Attached Figure Description

[0025] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0026] Figure 1 A flowchart illustrating a method for generating vehicle parking test cases provided in an embodiment of this application;

[0027] Figure 2 This is a schematic diagram of a process for generating multiple sets of target environment elements and element information corresponding to each frame of motion trajectory, as provided in an embodiment of this application.

[0028] Figure 3 This is a schematic diagram of a process for generating a target parking test scenario provided in an embodiment of this application;

[0029] Figure 4 A schematic flowchart of a vehicle parking test method provided in an embodiment of this application;

[0030] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0031] As mentioned earlier, assisted parking (also known as autonomous parking) is a crucial function in automotive intelligence. The parking algorithms involved in assisted parking systems are divided into two parts: perception algorithms and control algorithms. The trajectory planning and replanning functions of the control algorithms rely on the perception results output by the perception algorithms (such as parking spaces, static obstacles, curbs, and dynamic obstacles). In the early stages of project development, the perception algorithms are not yet mature, and the perception results are often inaccurate. Perception performance requires long-term iteration to stabilize. If the control algorithms are directly validated based on the environmental elements perceived by the perception algorithms in real-world driving scenarios, multiple real-world driving perception tests are required after the perception algorithm technology matures before parking control verification can be performed. This would severely slow down the development and iteration pace of parking control algorithms. Therefore, it is urgent to encapsulate parking simulation scenarios into parking test cases and input these test cases into the control algorithms before the perception algorithms mature, in order to verify the performance of the control algorithms in advance.

[0032] In existing technologies, to verify the performance of the planning and control algorithm in advance, a large number of parking simulation scenarios (also known as parking test scenarios) are created and input into the algorithm. Currently, there are many methods in the industry for constructing parking test scenarios, most of which involve arranging and combining existing environmental elements or adjusting some parameters to generate parking test scenarios. This method of combining collected environmental elements to generate a new parking test scenario, and then generalizing different parking test scenarios by changing the positions of the environmental elements or the vehicle's position, can only generate static parking test scenarios. Static parking test scenarios are relatively simple and singular, and can only verify the initial path planning capability of the planning and control algorithm in the parking scenario. It cannot verify the dynamic trajectory planning and replanning capability of the planning and control algorithm when there are dynamic obstacles or when obstacles change abruptly during vehicle movement.

[0033] Therefore, existing vehicle parking testing methods suffer from the problem that the generated parking simulation scenarios are simple and static, resulting in simple and incomplete parking test cases, which in turn leads to incomplete vehicle parking testing and incomplete vehicle parking function testing.

[0034] Based on this, this application provides a vehicle parking test case generation, parking test method, and computer device. By determining multiple environmental elements perceived by the vehicle in different parking scenarios and the element information of each environmental element, multiple sets of target environmental elements and their element information are generated corresponding to each frame of the vehicle's target motion trajectory. Multiple target parking test scenarios are generated based on these multiple sets of target environmental elements and their element information for each frame of the motion trajectory. Thus, multiple parking test scenarios can be generated based on the vehicle's motion trajectory, with environmental elements changing according to the vehicle's motion trajectory. In other words, the parking test scenarios are dynamically generated based on the vehicle's target motion trajectory. Furthermore, multiple parking test cases are generated from these multiple target parking test scenarios for vehicle parking testing. Therefore, multiple parking test scenarios can be dynamically generated along a single motion trajectory, increasing the complexity and diversity of the parking test scenarios, thereby increasing the complexity and diversity of the generated parking test cases. This makes vehicle parking testing more comprehensive, enabling testing of multiple vehicle parking functions as much as possible, and making vehicle parking function testing more complete.

[0035] Next, referring to the accompanying drawings, a detailed description will be given of the vehicle parking test case generation, parking test method, and computer equipment proposed in this application.

[0036] This application proposes a method for generating vehicle parking test cases, such as... Figure 1As shown, the method for generating vehicle parking test cases specifically includes the following steps.

[0037] S100 determines multiple environmental elements perceived by the vehicle in different parking scenarios, as well as element information of each environmental element.

[0038] In one implementation of this application, determining multiple environmental elements perceived by the vehicle in different parking scenarios and the element information of each environmental element includes: determining multiple driving image data of the vehicle under different driving attributes in different parking scenarios, obtaining multiple driving image data, the driving attributes including driving position, driving angle and driving pose, and performing perception processing on the multiple driving image data based on a perception algorithm to obtain multiple environmental elements of the vehicle in different parking scenarios and the element information of each environmental element, the element information of each environmental element including multiple element information corresponding to different driving attributes.

[0039] In the implementation of this application, the environmental elements include static environmental elements and dynamic environmental elements. The element information of the environmental elements includes the attribute information, location information, and relative position information of each environmental element and the vehicle.

[0040] Static environmental elements include parking spaces, pillars, road surfaces, walls, etc., while dynamic environmental elements include pedestrians, animals, vehicles, etc. The attribute information of environmental elements includes type, color, shape, etc.

[0041] Location information of environmental elements includes, for example, parking space coordinates, obstacle coordinates, and vehicle initial position information.

[0042] Among them, parking space coordinate information is used to define the geometric parameters of the parking target area, obstacle coordinate information is used to identify the position and outline of all obstacles (such as pillars, road rocks, walls, pedestrians, animals, and vehicles) in the parking scene, and vehicle initial position information is used to record the initial pose of the vehicle in the corresponding parking scene, i.e., coordinates. And the heading angle (yaw).

[0043] In this application, elements required in different parking scenarios are collected by real vehicles, such as pillars, pedestrians, road rocks, walls, and parking spaces. Since the purpose of the collection is to generate a large number of dynamic scenes later, the vehicle will collect the same element from different driving positions, different driving angles, different driving postures, and different driving distances, so that the perceived environmental elements include different position information, relative position and other element information due to the different distances and angles between the vehicle and the corresponding environmental elements.

[0044] Furthermore, the collected environmental elements and the results of each environmental element are stored in a structured format (such as txt, JSON or XML file) to build a standardized original scene data template, thus obtaining an environmental element database.

[0045] It should be noted that in this application, for the same type of environmental element, we will collect as much element information as possible from different scenarios. For example, there are many types of pillars, many colors, and different shapes. Therefore, we will try to cover the attribute information of different environmental elements.

[0046] S200: Determine the target motion trajectory of the vehicle. Based on multiple environmental elements and the element information of each environmental element, generate multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to each frame of the target motion trajectory.

[0047] As mentioned earlier, existing parking test scenario generation methods generate relatively simple scenarios that cannot simulate the jerking in perception and detection, nor can they simulate the dynamic effects of environmental elements.

[0048] Based on this, in this application, environmental elements corresponding to each frame of the vehicle's motion trajectory are dynamically generated according to the vehicle's motion trajectory. Furthermore, the field of view of the target camera and the perception jump simulation are performed on the environmental elements corresponding to each frame of the motion trajectory to make the generated parking test scenario more consistent with the real driving scenario of the vehicle, thus making the parking test scenario more diverse and realistic.

[0049] First, determine the target trajectory of the vehicle.

[0050] For example, the vehicle's movement trajectory in a real parking scenario can be obtained as the target trajectory. Alternatively, the vehicle's target trajectory can be generated randomly.

[0051] The target motion trajectory consists of T frame trajectory points. Each frame motion trajectory includes corresponding trajectory attributes, such as the vehicle's position information at the corresponding trajectory point and the vehicle's attitude information at the corresponding trajectory point (such as heading angle).

[0052] In this application, such as Figure 2 As shown, based on multiple environmental elements and the element information of each environmental element, multiple sets of target environmental elements and the element information of each set of target environmental elements are generated for each frame of the motion trajectory along the target motion trajectory, including the following steps.

[0053] S210, based on multiple environmental elements and the element information of each environmental element, as well as the trajectory attributes of the first frame motion trajectory in the target motion trajectory, generate multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to the first frame motion trajectory.

[0054] For example, based on multiple environmental elements and the element information of each environmental element, and considering the trajectory attributes of the first frame motion trajectory in the target motion trajectory, multiple sets of target environmental elements and the element information of each set of target environmental elements are randomly generated corresponding to the trajectory attributes (such as corresponding position and attitude) of the first frame motion trajectory.

[0055] S220, based on multiple environmental elements and the element information of each environmental element, multiple sets of target environmental elements corresponding to the (k-1)th frame motion trajectory in the target motion trajectory and the element information of each set of target environmental elements, and the trajectory attributes of the kth frame motion trajectory in the target motion trajectory, generate multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to the kth frame motion trajectory, where k ranges from 2 to T, and T is the total number of frames of the motion trajectory included in the target motion trajectory.

[0056] For example, from frame 2 to frame T, multiple environmental elements and their element information are generated sequentially, taking into account multiple sets of target environmental elements and their element information corresponding to the motion trajectory of frame k-1, as well as the trajectory attributes of the motion trajectory of frame k. In this way, based on the environmental elements corresponding to the motion trajectory of the previous frame and considering the trajectory attributes of the motion trajectory of the current frame, reasonable and coherent environmental elements that change with the motion trajectory are generated, so that the final generated parking test scenario is dynamic, reasonable and consistent with the vehicle's motion trajectory.

[0057] Furthermore, in one implementation of this application, based on the dynamic environment elements and element information of the multiple sets of target environment elements corresponding to the motion trajectory of the (k-1)th frame, and the trajectory attributes of the motion trajectory of the kth frame, the dynamic environment elements and element information of the multiple sets of target environment elements corresponding to the motion trajectory of the (k-1)th frame are updated, and multiple sets of target dynamic environment elements and element information of each set of target dynamic environment elements corresponding to the motion trajectory of the kth frame are generated.

[0058] For example, as the vehicle moves along its trajectory, the dynamic environmental elements (such as pedestrians and vehicles) in the parking test scenario may also change. This is necessary to test the trajectory replanning capability of the planning and control algorithm. Therefore, it is necessary to generate the motion trajectory of the dynamic environmental elements along with the vehicle's motion trajectory. In this application, based on the dynamic environmental elements corresponding to the motion trajectory of the previous frame and their element information, and combined with the trajectory attributes of the motion trajectory of the current frame, the element information of the dynamic environmental elements corresponding to the motion trajectory of the previous frame is updated (e.g., the position is updated) to generate multiple sets of target dynamic environmental elements corresponding to the motion trajectory of the current frame and the element information of each set of target dynamic environmental elements.

[0059] In another implementation of this application, multiple sets of target dynamic environment elements and element information of each set of target dynamic environment elements are generated based on the dynamic environment elements included in multiple environmental elements and the element information of each dynamic environment element, as well as the trajectory attributes of the motion trajectory of the k-th frame.

[0060] For example, the perception results of the perception algorithm cannot be guaranteed to be completely accurate, and dynamic environmental elements may exist in the previous frame and disappear in the next frame. Therefore, in order to test the trajectory replanning capability of the control algorithm, multiple sets of target dynamic environmental elements and element information of each set of target dynamic environmental elements can be generated based on the dynamic environmental elements included in multiple environmental elements and the element information of each dynamic environmental element, combined with the trajectory attributes of the motion trajectory of the k-th frame.

[0061] Thus, the multiple sets of target dynamic environment elements corresponding to the motion trajectory of the k-th frame may only include the dynamic environment elements corresponding to the motion trajectory of the (k-1)-th frame, or they may only include the dynamic environment elements from the multiple perceived environment elements. Alternatively, they may include both the dynamic environment elements corresponding to the motion trajectory of the (k-1)-th frame and other dynamic environment elements from the multiple environment elements. This allows us to obtain the motion trajectories of different dynamic environment elements as the vehicle's motion trajectory changes, enabling us to obtain multiple different parking test scenarios and increasing the complexity and diversity of parking test scenarios.

[0062] Furthermore, based on the static environmental elements included in multiple environmental elements and the element information of each static environmental element, the static environmental elements included in multiple target environmental elements corresponding to the motion trajectory of the (k-1)th frame and the element information of each static environmental element, and the trajectory attributes of the motion trajectory of the kth frame, multiple sets of target static environmental elements and the element information of each set of target static environmental elements corresponding to the motion trajectory of the kth frame are generated.

[0063] For example, for static environment elements, the static environment element corresponding to the motion trajectory of the current frame can be related to the target static environment element corresponding to the motion trajectory of the previous frame, or it can be unrelated. The target static environment element corresponding to the motion trajectory of the current frame can be regenerated from the static environment elements included by multiple environment elements to improve the variability of environment elements.

[0064] That is, in this application, the multiple sets of target static environment elements corresponding to the motion trajectory of the kth frame include the static environment elements included in the multiple sets of target environment elements corresponding to the motion trajectory of the (k-1)th frame, and / or, the static environment elements included in the multiple environment elements.

[0065] It should be noted that this application considers the trajectory attributes of the vehicle's target motion trajectory and can dynamically adjust the posture of the target environmental elements corresponding to each frame of the motion trajectory in order to better simulate the vehicle's perception effect and improve the realism of the parking test scenario.

[0066] Furthermore, based on the multiple sets of target dynamic environment elements corresponding to the motion trajectory of the kth frame and the element information of each set of target dynamic environment elements, as well as the multiple sets of target static environment elements corresponding to the motion trajectory of the kth frame and the element information of each set of target static environment elements, the multiple sets of target environment elements corresponding to the motion trajectory of the kth frame and the element information of each set of target environment elements are obtained.

[0067] Thus, the multiple sets of target dynamic environment elements corresponding to the motion trajectory of the kth frame may include both dynamic environment elements and static environment elements. The dynamic environment elements and static environment elements can be target environment elements related to the motion trajectory of the (k-1)th frame, or they can be new environment elements. This makes the multiple sets of target dynamic environment elements corresponding to the motion trajectory of each frame more diverse.

[0068] In the implementation of this application, a scene constraint graph is also constructed. Based on the scene constraint graph, the method described above is used to determine multiple sets of target environment elements corresponding to the motion trajectory of the k-th frame and the element information of each set of target environment elements, so as to constrain the multiple sets of target environment elements corresponding to the motion trajectory of the k-th frame and the element information of each set of target environment elements within the scope of the scene constraint graph.

[0069] The constraint types in the scene constraint graph include, but are not limited to, collision constraints, minimum lane width, minimum turning radius of the vehicle, and parking space size.

[0070] It should also be noted that this application generates a vehicle's trajectory that can reach the designated parking space by determining the target trajectory. Therefore, it can generate a vehicle parking test scenario corresponding to the complete parking trajectory of the vehicle, which is convenient for subsequent testing of whether the control algorithm can completely simulate the vehicle's movement to the target parking space based on the parking test scenario.

[0071] S300 generates multiple target parking test scenarios based on multiple sets of target environment elements corresponding to each frame's motion trajectory and the element information of each set of target environment elements.

[0072] For example, based on multiple sets of target environment elements corresponding to multiple frames of motion trajectories and the element information of each set of target environment elements, multiple target parking test scenarios containing the complete target motion trajectory of the vehicle are generated, that is, one target motion trajectory corresponds to multiple target parking test scenarios.

[0073] In this application, such as Figure 3As shown, multiple target parking test scenarios are generated based on multiple sets of target environment elements corresponding to each frame's motion trajectory and the element information of each set of target environment elements, including the following steps.

[0074] S310, determine the multiple sets of target environment elements within the field of view of the target camera corresponding to the vehicle among the multiple sets of target environment elements corresponding to the motion trajectory of each frame, and obtain the multiple sets of visible target environment elements corresponding to the motion trajectory of each frame.

[0075] For example, not all of the multiple sets of target environment elements corresponding to each frame of motion trajectory are input to the planning and control algorithm. This application simulates the field of view (FOV) effect of a camera, so that the scene information corresponding to each frame of motion trajectory in the parking test scenario input to the planning and control algorithm is multiple sets of visible target environment elements that are more in line with the field of view of the vehicle's target camera. Target environment elements that are not in the field of view of the vehicle's target camera will not be input to the planning and control algorithm, thereby reducing the interference of useless information and making the vehicle parking test more in line with the requirements of the real vehicle test scenario.

[0076] Among them, the camera field of view is the field of view of the camera corresponding to the vehicle under test, or the field of view of the camera corresponding to the vehicle's front-view camera, rear-view camera, or 360-degree camera with a common field of view.

[0077] S320 performs perception jump simulation on the element information of multiple sets of visible target environment elements corresponding to each frame motion trajectory, and generates target element information of multiple sets of visible target environment elements corresponding to each frame motion trajectory.

[0078] For example, in the actual perception process of a vehicle, there will inevitably be perception noise, position jitter, confidence fluctuations, etc. Therefore, in order to make the parking test scenario more in line with the detection jump characteristics of real perception, perception jump simulation is performed on the element information of multiple sets of visible target environment elements corresponding to each frame of motion trajectory. Reasonable noise, position jitter, and confidence fluctuation are introduced into the element information of visible target environment elements that need to be input to the control algorithm, so as to achieve a more realistic perception simulation effect.

[0079] S330 generates multiple target parking test scenarios based on multiple sets of visible target environment elements corresponding to each frame's motion trajectory and the target element information corresponding to each set of visible target environment elements.

[0080] For example, multiple sets of visible target environment elements corresponding to each frame's motion trajectory, as well as the target element information corresponding to each set of visible target environment elements, are encapsulated to generate multiple target parking test scenarios.

[0081] The system encapsulates multiple sets of visible target environment elements corresponding to multi-frame motion trajectories, as well as each set of visible target environment elements, and determines feasible solutions for target motion trajectories to ensure that the vehicle can reach the parking space.

[0082] Furthermore, in another implementation of this application, a target parking test scenario can also be generated based on a scenario generation model.

[0083] For example, the scene generation requirements are determined, and the scene generation requirements, multiple environmental elements and element information of each environmental element and the vehicle's motion trajectory are input into the scene generation model so that the scene generation model can determine the vehicle's target motion trajectory. Based on the multiple environmental elements and element information of each environmental element, multiple sets of target environmental elements and element information of each set of target environmental elements are generated for each frame of the target motion trajectory. Based on the multiple sets of target environmental elements and element information of each set of target environmental elements for each frame of the motion trajectory, multiple target parking test scenarios corresponding to the scene generation requirements are generated.

[0084] Specifically, the dynamic scene generalization tool integrates a general large language model (as an example of a scene generation model). Users describe the required scene, generate scene generation requirements, and input the scene generation requirements, multiple environmental elements, element information of each environmental element, and the motion trajectory perceived by the actual vehicle into the scene generation model. The scene generation model can then determine the target motion trajectory of the vehicle, extract multiple environmental elements and element information of each environmental element from the environmental element database, and automatically generate a large number of simulation test scenarios.

[0085] For example, if the input scenario generation requirement is "generate a parking lot replanning scenario where a pedestrian crosses the parking lot", the scenario generation model can generate parking test scenarios where a pedestrian crosses the parking lot from various angles.

[0086] The S400 generates multiple parking test cases based on multiple target parking test scenarios for vehicle parking testing.

[0087] For example, the generalized parking test scenarios are encapsulated into multiple parking test cases that can be executed independently, reproduced, and managed in batches, forming a large-scale dynamic scenario library for vehicle parking testing.

[0088] Furthermore, this application also provides a vehicle parking test method, such as... Figure 4 As shown, the specific steps include the following.

[0089] S500 identifies multiple parking test cases.

[0090] Each parking test case is generated based on the aforementioned parking test case generation method.

[0091] The S600 performs vehicle parking tests based on various parking test cases.

[0092] In this application, vehicle parking tests are performed based on each parking test case, including: encapsulating each parking test case into a corresponding test container, the test container including a rule control algorithm, and injecting the corresponding parking test case into the corresponding rule control algorithm so that the rule control algorithm simulates vehicle parking to achieve vehicle parking test.

[0093] For example, multiple parking test cases are encapsulated into multiple independent test containers according to preset encapsulation rules (such as scenario complexity and parking type). The encapsulated rule control algorithms are run in parallel by multiple containers to run multiple parking test cases in parallel and perform synchronous testing of multiple parking test scenarios, thereby improving parking test efficiency.

[0094] Furthermore, traditional methods only increase the number of containers, resulting in low hardware utilization. In this application, the simulation time within the test container is accelerated at the code level, increasing the frame rate of the parking test scenario with injected algorithms from the usual 20 frames / second to 100 frames / second or more. Without increasing the number of containers, hardware utilization and simulation speed are significantly improved, enabling rapid batch verification of large-scale scenarios.

[0095] The parking test cases based on this application can verify the initial path planning capability of the planning algorithm, the replanning capability after dynamic obstacle intrusion, the trajectory tracking and low-level control capability, noise perception, robustness under jump conditions, and the ability of the vehicle to reliably park in the target parking space.

[0096] This application also automatically records the test results for each test case, including indicators such as planning success rate, path smoothness, collision detection, and time consumption, thereby enabling a comprehensive and efficient evaluation of the performance and robustness of the parking planning algorithm.

[0097] The parking test case generation method and vehicle parking test method of this application are essentially a method for generating a large-scale scene library and conducting parallel testing for autonomous driving parking planning. Unlike mainstream static scenes in the industry, this application can generate dynamic parking scenes with continuous temporal sequence, movable targets, and variable states, covering real-world conditions such as replanning and dynamic obstacle avoidance. Furthermore, it can simulate perception detection jumps, camera FOV occlusion, and distance / angle-related perception deviations, more closely resembling real-world perception output and supporting regulation and control optimization based on real-world perception performance. Further, this application integrates a large language model, supporting the direct generation of a large number of target scenes through text descriptions, lowering the threshold for scene construction and improving scene coverage efficiency. By accelerating test container time, it significantly improves simulation testing efficiency, greatly increases hardware utilization and test throughput, meets the rapid iterative testing needs of large-scale scene libraries, and can completely verify the entire parking regulation and control process from initial planning and dynamic correction to final parking.

[0098] This application also provides a computer device for executing a vehicle parking test case generation method and / or a parking test method.

[0099] Please see Figure 5 , Figure 5 The diagram shown is a structural schematic of a computer device provided in an embodiment of this application. Figure 5 As shown, the computer device may include: transceiver 121, processor 122, and memory 123.

[0100] Processor 122 executes computer execution instructions stored in memory, causing it to execute the vehicle parking test case generation method and / or parking test method described in the above embodiments. Processor 122 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0101] The memory 123 is connected to the processor 122 via the system bus and completes communication between them. The memory 123 is used to store computer program instructions.

[0102] In this application, the computer equipment may be, for example, a computer, an industrial control computer, a remote terminal, or a cloud server, and may also be an on-board computer device in a vehicle.

[0103] This application also provides a chip for executing instructions, which is used to execute the vehicle parking test case generation method and / or parking test method technical solutions in the above embodiments.

[0104] This application also provides a computer-readable storage medium storing computer instructions. When the computer instructions are executed on the processor of a computer device, the processor of the computer device executes the vehicle parking test case generation method and / or the technical solution of the parking test method described in the above embodiments.

[0105] This application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When the at least one processor executes the computer program, it can implement the technical solutions of the vehicle parking test case generation method and / or parking test method described in the above embodiments.

[0106] It should be noted that, in addition to the specific embodiments described above, those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Although the description of this application is presented in conjunction with preferred embodiments, this does not mean that the features of this application are limited to this implementation. On the contrary, the purpose of describing the application in conjunction with the implementation is to cover other options or modifications that may be derived from this application. To provide a thorough understanding of this application, many specific details are included in the above description, and this application may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of this application, some specific details will be omitted in the description. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0107] It should be noted that in this specification, similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0108] It should be noted that some structural or methodological features may be shown in the accompanying drawings in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be necessary. Rather, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the illustrative drawings. Furthermore, including structural or methodological features in a particular figure does not imply that such features are required in all embodiments, and in some embodiments, these features may be omitted or may be combined with other features.

[0109] Although this application has been illustrated and described with reference to certain preferred embodiments, those skilled in the art should understand that the above description is a further detailed explanation of the application in conjunction with specific implementations, and should not be construed as limiting the specific implementation of the application to these descriptions. Those skilled in the art can make various changes in form and detail, including some simple deductions or substitutions, without departing from the spirit and scope of this application.

Claims

1. A method for generating vehicle parking test cases, characterized in that, The method includes: Determine multiple environmental elements perceived by the vehicle in different parking scenarios, as well as the element information of each environmental element; The target motion trajectory of the vehicle is determined, and based on the multiple environmental elements and the element information of each environmental element, multiple sets of target environmental elements and the element information of each set of target environmental elements are generated for each frame of the motion trajectory in the target motion trajectory. Based on the multiple sets of target environment elements corresponding to each frame of motion trajectory and the element information of each set of target environment elements, multiple target parking test scenarios are generated. Based on the multiple target parking test scenarios, multiple parking test cases are generated for vehicle parking testing.

2. The method for generating vehicle parking test cases according to claim 1, characterized in that, Based on the plurality of environmental elements and the element information of each environmental element, multiple sets of target environmental elements and the element information of each set of target environmental elements are generated for each frame of the target motion trajectory, including: Based on the plurality of environmental elements and the element information of each environmental element, and the trajectory attributes of the first frame motion trajectory in the target motion trajectory, generate multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to the first frame motion trajectory. Based on the multiple environmental elements and the element information of each environmental element, the multiple sets of target environmental elements corresponding to the (k-1)th frame motion trajectory in the target motion trajectory and the element information of each set of target environmental elements, and the trajectory attributes of the kth frame motion trajectory in the target motion trajectory, multiple sets of target environmental elements and the element information of each set of target environmental elements corresponding to the kth frame motion trajectory are generated, where k ranges from 2 to T, and T is the total number of frames of the motion trajectory included in the target motion trajectory.

3. The method for generating vehicle parking test cases according to claim 2, characterized in that, Based on the plurality of environmental elements and the element information of each environmental element, the plurality of target environmental elements corresponding to the (k-1)th frame motion trajectory in the target motion trajectory and the element information of each group of target environmental elements, and the trajectory attributes of the kth frame motion trajectory in the target motion trajectory, the plurality of target environmental elements corresponding to the kth frame motion trajectory and the element information of each group of target environmental elements are generated, including: Based on the dynamic environment elements and element information of the multiple sets of target environment elements corresponding to the (k-1)th frame motion trajectory, and the trajectory attributes of the k-th frame motion trajectory, update the dynamic environment elements and element information of the multiple sets of target environment elements corresponding to the (k-1)th frame motion trajectory, generate multiple sets of target dynamic environment elements and element information of each set of target dynamic environment elements corresponding to the k-th frame motion trajectory, and / or based on the dynamic environment elements and element information of each set of target dynamic environment elements included in the multiple environment elements, and the trajectory attributes of the k-th frame motion trajectory, generate multiple sets of target dynamic environment elements and element information of each set of target dynamic environment elements corresponding to the k-th frame motion trajectory; Based on the static environmental elements included in the plurality of environmental elements and the element information of each static environmental element, the static environmental elements included in the plurality of target environmental elements corresponding to the motion trajectory of the (k-1)th frame and the element information of the static environmental elements, and the trajectory attributes of the motion trajectory of the kth frame, the plurality of target static environmental elements corresponding to the motion trajectory of the kth frame and the element information of each target static environmental element are generated. Based on the multiple sets of target dynamic environment elements corresponding to the motion trajectory of the kth frame and the element information of each set of target dynamic environment elements, as well as the multiple sets of target static environment elements corresponding to the motion trajectory of the kth frame and the element information of each set of target static environment elements, the multiple sets of target environment elements corresponding to the motion trajectory of the kth frame and the element information of each set of target environment elements are obtained.

4. The method for generating vehicle parking test cases according to claim 3, characterized in that, Based on the multiple sets of target environment elements corresponding to each frame's motion trajectory and the element information of each set of target environment elements, multiple target parking test scenarios are generated, including: Determine the multiple sets of target environment elements that are within the field of view of the target camera corresponding to the vehicle among the multiple sets of target environment elements corresponding to each frame of motion trajectory, and obtain the multiple sets of visible target environment elements corresponding to each frame of motion trajectory. A perceptual jump simulation is performed on the element information of the multiple sets of visible target environment elements corresponding to each frame motion trajectory to generate target element information of the multiple sets of visible target environment elements corresponding to each frame motion trajectory; Based on the multiple sets of visible target environment elements corresponding to each frame of motion trajectory and the target element information corresponding to each set of visible target environment elements, multiple target parking test scenarios are generated.

5. The method for generating vehicle parking test cases according to claim 1, characterized in that, The method further includes generating the target parking test scenario in the following manner: Determine the scene generation requirements, input the scene generation requirements, the multiple environmental elements and their element information, and the vehicle's motion trajectory into the scene generation model, so that the scene generation model determines the vehicle's target motion trajectory. Based on the multiple environmental elements and their element information, generate multiple sets of target environmental elements and their element information corresponding to each frame of the target motion trajectory. Based on the multiple sets of target environmental elements and their element information corresponding to each frame of the motion trajectory, generate multiple target parking test scenarios corresponding to the scene generation requirements.

6. The method for generating vehicle parking test cases according to any one of claims 1-5, characterized in that, Determine multiple environmental elements perceived by the vehicle in different parking scenarios, as well as the element information of each environmental element, including: Multiple driving image data of a vehicle under different driving attributes in different parking scenarios are determined to obtain multiple driving image data. The driving attributes include driving position, driving angle and driving posture. Based on the perception algorithm, the multiple driving image data are processed to obtain multiple environmental elements of the vehicle in different parking scenarios and the element information of each environmental element. The element information of each environmental element includes multiple element information corresponding to the different driving attributes.

7. The method for generating vehicle parking test cases according to claim 6, characterized in that, The element information of the environmental element includes the attribute information, location information, and relative position information of each environmental element to the vehicle.

8. A vehicle parking test method, characterized in that, The method includes: Multiple parking test cases are determined, wherein the multiple parking test cases are generated based on the vehicle parking test case generation method according to any one of claims 1-7; The vehicle parking test is performed based on each of the parking test cases.

9. The vehicle parking test method according to claim 8, characterized in that, Based on the parking test cases described above, vehicle parking tests are conducted, including: Each parking test case is encapsulated into a corresponding test container, and the test container includes a control algorithm; At least one of the test containers injects the corresponding parking test cases into the corresponding regulation algorithm, so that the regulation algorithm simulates the vehicle parking to achieve vehicle parking test.

10. A computer device, characterized in that, The computer device includes: a processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes the computer execution instructions stored in the memory to enable the computer device to implement the vehicle parking test case generation method as described in any one of claims 1-7, and / or the vehicle parking test method as described in any one of claims 8-9.

Images