Vehicle simulation task processing method, device, equipment, medium and program product

Abstract

Embodiments of the present application provide a vehicle simulation task processing method, device, equipment, medium and program product, scene configuration content and a target simulation task are obtained; the scene configuration content and the target simulation task are uploaded to the cloud end, so that the cloud end creates a target scene model according to the scene configuration content, and at least one target automatic driving algorithm is determined from the automatic driving algorithm according to the target simulation task; the target scene model and the target automatic driving algorithm in the cloud end are obtained; according to the target simulation task, a target vehicle model object is determined from a vehicle model object; the target vehicle model object is controlled to simulate in the target scene model by using the target automatic driving algorithm, and target simulation data is obtained. Embodiments of the present application form a full-automatic simulation process of the automatic driving algorithm, effectively adapt to the verification needs of complex automatic driving algorithms, support large-scale concurrent simulation experiments, and significantly reduce the dependence cost of automatic driving algorithm simulation on entity vehicle hardware.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to methods, apparatus, equipment, media, and program products for processing vehicle simulation tasks. Background Technology

[0002] The autonomous driving simulation platform technology solutions in related technologies mainly exhibit two major technical characteristics: First, based on offline-built three-dimensional (3D) scene assets, scene generation and management are achieved through manual arrangement, and testing and verification are carried out in a cloud simulation environment; second, a hardware-in-the-loop simulation mode is adopted, which connects to the real vehicle hardware system through video signal injection to build a closed testing environment.

[0003] However, the aforementioned autonomous driving simulation platforms still have significant limitations. First, they lack the ability to adapt to complex autonomous driving algorithms and cannot provide a large-scale concurrent algorithm verification environment, thus restricting the system iteration efficiency for mass production. Second, scene construction relies on manual editing and limited pre-set scene assets, resulting in insufficient scene diversity and low realism, affecting the credibility of simulation results. Third, hardware-in-the-loop solutions require the configuration of complete vehicle-side hardware, leading to high testing costs and difficulty in scaling up, while traditional cloud simulation solutions have technical bottlenecks in data synchronization efficiency and resource scheduling optimization, making it difficult to support high-density concurrent demands.

[0004] Therefore, how to further optimize existing autonomous driving simulation platforms is an urgent problem to be solved in this field. Summary of the Invention

[0005] This application provides a method for processing vehicle simulation tasks to solve or at least partially solve the above-mentioned technical problems.

[0006] Accordingly, embodiments of this application also provide a vehicle simulation task processing device, an electronic device, and a storage medium to ensure the implementation and application of the above method.

[0007] To address the aforementioned problems, this application discloses a method for processing vehicle simulation tasks, applied to an automotive simulation platform. The automotive simulation platform includes vehicle model objects and is connected to a cloud computing center, where an autonomous driving algorithm is stored. The method includes:

[0008] Obtain the scene configuration content and target simulation task;

[0009] The scenario configuration content and the target simulation task are uploaded to the cloud so that the cloud can create a target scenario model based on the scenario configuration content and determine at least one target autonomous driving algorithm from the autonomous driving algorithms based on the target simulation task.

[0010] Obtain the target scene model and the target autonomous driving algorithm from the cloud;

[0011] Based on the target simulation task, determine the target vehicle model object from the vehicle model object;

[0012] The target vehicle model object is controlled to perform simulation using the target autonomous driving algorithm in the target scene model to obtain target simulation data.

[0013] Optionally, the scene configuration content includes scene configuration data and scene configuration tasks, the vehicle simulation platform includes task nodes, the vehicle simulation platform is communicatively connected to the actual vehicle, and obtaining the scene configuration content includes:

[0014] At least two of the task nodes are combined using a directed acyclic graph method to obtain a scene configuration task, which is a task used to construct the target scene model, and there are upstream-downstream relationships among the task nodes in the scene configuration task;

[0015] Obtain the scene configuration data collected from the actual vehicle, wherein the scene configuration data is scene data used to construct the target scene model;

[0016] The scene configuration content is obtained based on the scene configuration data and the scene configuration task.

[0017] Optionally, the scene configuration task includes a data annotation task and a scene construction task, wherein the data annotation task and the scene construction task each have corresponding task nodes, and uploading the scene configuration content and the target simulation task to the cloud includes:

[0018] The scene configuration data and the data annotation task are uploaded to the cloud so that the cloud can annotate the scene configuration data;

[0019] Obtain the node status of the task nodes for the data annotation task;

[0020] When the node status is "completed", retrieve the labeled scene configuration data from the cloud.

[0021] Based on the labeled scene configuration data, a task start command is sent to the cloud, which is used to cause the cloud to start executing the scene construction task.

[0022] Optionally, the method further includes:

[0023] Obtain the current task node and task status of the scene configuration task in the cloud. The current task node is the task node currently being processed in the scene configuration task. There are downstream task nodes of the current task node in the task nodes of the scene configuration task.

[0024] When the task status is "completed", the execution result of the current task node is obtained from the cloud.

[0025] Based on the execution result, a node start command is sent to the cloud, which is used to enable the cloud to start processing the downstream task node.

[0026] Optionally, the vehicle simulation platform includes a cache area, and the step of acquiring the target scene model and the target autonomous driving algorithm from the cloud includes:

[0027] When the cache does not store the target scene model and / or the target autonomous driving algorithm, the target scene model and / or the target autonomous driving algorithm in the cloud are obtained.

[0028] The method further includes:

[0029] When the cache contains the target scene model and / or the target autonomous driving algorithm, the target scene model and / or the target autonomous driving algorithm in the cache are retrieved.

[0030] Optionally, the vehicle simulation platform includes a temporary storage area, and after acquiring the target scene model and / or the target autonomous driving algorithm from the cloud, the method includes:

[0031] Store the target scene model and / or the target autonomous driving algorithm in the temporary storage area;

[0032] Verify the target scene model and / or the target autonomous driving algorithm;

[0033] When the target scene model and / or the target autonomous driving algorithm pass the verification, the target scene model and / or the target autonomous driving algorithm are copied to the cache area.

[0034] Optionally, the target autonomous driving algorithm in the cache is associated with a first usage frequency, and the target scene model in the cache is associated with a second usage frequency. The method further includes:

[0035] When acquiring the target scene model and / or the target autonomous driving algorithm in the cache, update the first usage frequency and / or the second usage frequency;

[0036] Obtain the data storage capacity and storage limit of the cache area;

[0037] When the amount of data stored exceeds the storage limit, data to be deleted is determined from the target autonomous driving algorithm and the target scene model in the cache based on the first usage frequency and the second usage frequency.

[0038] Delete the data to be deleted.

[0039] Optionally, after controlling the target vehicle model object to perform simulation using the target autonomous driving algorithm in the target scene model to obtain target simulation data, the method further includes:

[0040] The target autonomous driving algorithm is evaluated based on the target simulation data to obtain evaluation data corresponding to the target autonomous driving algorithm, and the evaluation data represents the performance level of the target autonomous driving algorithm.

[0041] This application also discloses a vehicle simulation task processing device applied to an automotive simulation platform. The automotive simulation platform includes vehicle model objects and is communicatively connected to a cloud. The cloud stores autonomous driving algorithms. The device includes:

[0042] The simulation configuration acquisition module is used to acquire scene configuration content and target simulation tasks;

[0043] The simulation scene creation module is used to upload the scene configuration content and the target simulation task to the cloud, so that the cloud can create a target scene model according to the scene configuration content and determine at least one target autonomous driving algorithm from the autonomous driving algorithms according to the target simulation task;

[0044] The cloud data acquisition module is used to acquire the target scene model and the target autonomous driving algorithm in the cloud.

[0045] The simulation object acquisition module is used to determine the target vehicle model object from the vehicle model objects according to the target simulation task;

[0046] The simulation task execution module is used to control the target vehicle model object to perform simulation in the target scene model using the target autonomous driving algorithm to obtain target simulation data.

[0047] This application also discloses an electronic device, including: a processor; and a memory storing executable code thereon, which, when executed, causes the processor to perform a processing method for one or more vehicle simulation tasks as described in the embodiments of this application.

[0048] This application also discloses one or more machine-readable media storing executable code thereon, which, when executed, causes a processor to perform a processing method for one or more vehicle simulation tasks as described in this application.

[0049] This application also discloses a computer program product, including a computer program / instruction, which, when executed by a processor, implements a processing method for one or more vehicle simulation tasks as described in the embodiments of this application.

[0050] Compared with the prior art, the embodiments of this application have the following advantages:

[0051] In this embodiment, the vehicle simulation platform acquires scene configuration content and target simulation tasks; uploads the scene configuration content and target simulation tasks to the cloud, so that the cloud can create a target scene model based on the scene configuration content, and determine at least one target autonomous driving algorithm from autonomous driving algorithms based on the target simulation tasks; acquires the target scene model and target autonomous driving algorithm from the cloud; determines a target vehicle model object from vehicle model objects based on the target simulation tasks; and controls the target vehicle model object to perform simulation using the target autonomous driving algorithm in the target scene model to obtain target simulation data. This embodiment, by storing the autonomous driving algorithm in the cloud and automatically constructing the target scene model in the cloud, builds a stable and reliable cloud deployment environment for the vehicle simulation platform, realizing a complete data backhaul and scene reconstruction link. When a simulation task is triggered, the target scene model and target autonomous driving algorithm can be acquired from the cloud and sent to the vehicle simulation platform, driving the target vehicle model object to execute the simulation of the target autonomous driving algorithm in the target scene model and generating target simulation data, thereby forming a fully automated simulation process for autonomous driving algorithms, effectively adapting to the verification requirements of complex autonomous driving algorithms.

[0052] Compared to simulation platforms in related technologies that cannot provide a large-scale concurrent algorithm verification environment, the automotive simulation platform in this application embodiment can simultaneously select multiple target vehicle models for simulation, thereby supporting large-scale concurrent simulation experiments. It also significantly reduces the cost of relying on physical vehicle-mounted hardware for autonomous driving algorithm simulation in related technologies. Furthermore, this application embodiment constructs a highly realistic target scene model through scene configuration content, improving the mapping accuracy between the simulation environment and real roads, and solving the problems of insufficient scene diversity and low realism in simulation platforms of related technologies. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the steps of an embodiment of a vehicle simulation task processing method according to this application.

[0054] Figure 2This is a schematic diagram of the task node combination according to an embodiment of a vehicle simulation task processing method of this application;

[0055] Figure 3 This is a schematic diagram illustrating data downloading according to an embodiment of a vehicle simulation task processing method of this application;

[0056] Figure 4 This is a schematic diagram of an automotive simulation platform architecture according to an embodiment of a vehicle simulation task processing method of this application;

[0057] Figure 5 This is a structural block diagram of an embodiment of a vehicle simulation task processing device according to this application;

[0058] Figure 6 This is a schematic diagram of the structure of a device provided in an embodiment of this application. Detailed Implementation

[0059] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0060] The technical solutions for autonomous driving simulation platforms in related technologies mainly exhibit two major technical characteristics. The first is based on offline-built 3D scene assets, with scene generation and management achieved through manual orchestration, and testing and verification relying on a cloud simulation environment. Currently, autonomous driving technology is in a phase of rapid upgrading; however, this type of simulation platform still has significant shortcomings in simulating and verifying complex autonomous driving algorithms. It fails to effectively integrate complex, computationally intensive, and real-time-critical autonomous driving algorithms with simulation scene assets, resulting in a lack of a realistic deployment environment for complex algorithms. Therefore, they cannot provide large-scale, high-concurrency simulation verification support for the mass production and deployment of autonomous driving algorithms. Furthermore, existing platforms primarily target small-batch scene editing and the maintenance and management of limited assets. The scene sources are singular, mainly relying on manual construction, which inevitably differs from real-world scenarios, further limiting the realism and credibility of the simulation.

[0061] The second approach is to use hardware-in-the-loop (HIL) simulation, which involves injecting video signals into the actual vehicle hardware system to create a closed testing environment. This HIL simulation solution requires a complete set of autonomous driving vehicle hardware for each test, resulting in high costs for mass implementation. Therefore, this approach struggles to support large-scale, high-concurrency automated cloud simulation requirements. Furthermore, offline, manually constructed 3D scenes have significant limitations in scale and quantity, and also struggle to guarantee the realism and credibility of the simulation, further restricting its application value in autonomous driving algorithm verification.

[0062] In the development of autonomous driving technology, data-driven end-to-end large-scale models have become the mainstream trend for achieving advanced intelligent driving. The necessity of end-to-end data-driven methods lies not only in their ability to significantly reduce system complexity and improve efficiency, but also in their strong adaptability to dynamic and uncertain driving environments. Under complex and changing road conditions, data-driven deep learning models can process and respond to diverse driving scenarios in real time, thereby effectively improving driving safety and comfort. Furthermore, with the rapid development of big data technology and the continuous enhancement of computing power, end-to-end data-driven methods possess the ability to continuously learn and self-optimize, better addressing the increasingly complex traffic challenges of the future. Therefore, in the pursuit of safer, more efficient, and more cost-effective autonomous driving algorithms, end-to-end data-driven methods have become an indispensable key technological support. However, due to the shortcomings of the autonomous driving simulation platform solutions in the aforementioned related technologies, most autonomous driving simulation platforms do not support large-scale concurrent simulation of autonomous driving algorithms based on end-to-end data-driven methods.

[0063] To address this, this application proposes a method for processing vehicle simulation tasks. The vehicle simulation platform in this application completes a comprehensive data backhaul and scene construction chain, realizes large-scale cloud simulation based on scene models, and achieves end-to-end data-driven closed-loop chain functionality. This solves, or at least partially solves, the technical problem of how to further optimize existing autonomous driving simulation platforms to provide a large-scale concurrent algorithm verification environment and a highly realistic target scene model, and reduce the cost of autonomous driving algorithm simulation's dependence on physical vehicle hardware. This provides a more complete vehicle simulation platform for the simulation and optimization of autonomous driving algorithms.

[0064] Reference Figure 1 This is a flowchart illustrating the steps of an embodiment of a vehicle simulation task processing method according to this application, including the following steps:

[0065] Step 101: Obtain the scene configuration content and target simulation task.

[0066] The vehicle simulation task processing method described in this application is applied to an automotive simulation platform. The automotive simulation platform can be built based on a Graphics Processing Unit (GPU) cluster. Each GPU can be considered a test vehicle, i.e., a vehicle model object. Each vehicle model object corresponds to an independent autonomous driving system and a simulation engine that drives the autonomous driving algorithm and scene model. The automotive simulation platform can contain multiple vehicle model objects simultaneously. Compared to limited real-world vehicle resources, using a GPU cluster offers advantages such as lower cost, higher stability, and stronger generalization capabilities.

[0067] The vehicle simulation platform in this embodiment of the application is also connected to the cloud, where autonomous driving algorithms are stored. When the vehicle model object performs a vehicle simulation task, the corresponding autonomous driving algorithm is downloaded from the cloud. The autonomous driving algorithm may include complex autonomous driving algorithms such as end-to-end algorithms and visual language model (VLM) algorithms.

[0068] In step 101 of this embodiment, the scene configuration content and the target simulation task are obtained. The scene configuration content is used to enable the cloud to create a target scene model, and the target simulation task is the simulation task that the vehicle model object needs to perform subsequently. The target simulation task can be the verification of autonomous driving algorithms, specifically including the verification of perception algorithms (such as object detection, recognition, and tracking), the verification of decision planning algorithms (such as path planning and behavior decision-making), the verification of control algorithms (such as steering, throttle, and braking control), and the verification of system integration (such as module collaboration and overall performance testing).

[0069] Step 102: Upload the scene configuration content and the target simulation task to the cloud so that the cloud can create a target scene model based on the scene configuration content and determine at least one target autonomous driving algorithm from the autonomous driving algorithms based on the target simulation task.

[0070] In step 102, the scene configuration content and the target simulation task are uploaded to the cloud so that the cloud can automatically create a target scene model based on the scene configuration content, and determine the target autonomous driving algorithm from the autonomous driving algorithm based on the target simulation task.

[0071] The scene configuration content includes real scene configuration data collected from actual vehicles. The target scene model created based on this data can reconstruct large-scale world scene models, solving the problems of insufficient scene diversity and low realism caused by small-batch scene editing and management in related technologies. In this embodiment, the target autonomous driving algorithm can be one or more, and the type, version, and number of the target autonomous driving algorithm are determined according to the actual needs of the target simulation task.

[0072] Step 103: Obtain the target scene model and the target autonomous driving algorithm from the cloud.

[0073] In step 103, when the vehicle simulation platform needs to perform the target simulation task, the vehicle simulation platform automatically downloads the target scene model and target autonomous driving algorithm required for the target simulation task from the cloud.

[0074] Step 104: Based on the target simulation task, determine the target vehicle model object from the vehicle model objects.

[0075] In step 104, since the automotive simulation platform can contain multiple vehicle model objects simultaneously, before executing the target simulation task, it is necessary to determine the target vehicle model object from the vehicle model objects in the automotive simulation platform according to the target simulation task. In one embodiment, there can be multiple target vehicle model objects, so that multiple target vehicle model objects can execute different target autonomous driving algorithms in parallel, supporting large-scale concurrent simulation experiments and realizing large-scale cloud simulation based on target scene models.

[0076] Step 105: Control the target vehicle model object to perform simulation in the target scene model using the target autonomous driving algorithm to obtain target simulation data.

[0077] In step 105, within the vehicle simulation platform, the target vehicle model object is used to simulate the target autonomous driving algorithm within the target scene model. In one embodiment, the target simulation task can be set to automatically loop multiple times. That is, after the target vehicle model object completes the simulation using the target autonomous driving algorithm within the target scene model, the target vehicle model object, the target scene model, and the target autonomous driving algorithm are automatically reset to their initial states, and the simulation is repeated until the target simulation task loops to a preset number of times.

[0078] This application embodiment stores the autonomous driving algorithm in the cloud and automatically constructs the target scene model in the cloud, thus creating a stable and reliable cloud deployment environment for the automotive simulation platform and realizing a complete data backhaul and scene reconstruction link. When the simulation task is triggered, the target scene model and the target autonomous driving algorithm can be obtained from the cloud to the automotive simulation platform, driving the target vehicle model object to execute the simulation of the target autonomous driving algorithm in the target scene model and generating target simulation data, thereby forming a fully automatic simulation process for the autonomous driving algorithm, effectively adapting to the verification requirements of complex autonomous driving algorithms.

[0079] Compared to simulation platforms in related technologies that cannot provide a large-scale concurrent algorithm verification environment, the automotive simulation platform in this application embodiment can simultaneously select multiple target vehicle models for simulation, thereby supporting large-scale concurrent simulation experiments. It also significantly reduces the cost of relying on physical vehicle-mounted hardware for autonomous driving algorithm simulation in related technologies. Furthermore, this application embodiment constructs a highly realistic target scene model through scene configuration content, improving the mapping accuracy between the simulation environment and real roads, and solving the problems of insufficient scene diversity and low realism in simulation platforms of related technologies.

[0080] In one optional embodiment of this application, the scene configuration content includes scene configuration data and scene configuration tasks, the car simulation platform includes task nodes, the car simulation platform is communicatively connected to the actual vehicle, and step 101 may further include:

[0081] At least two of the task nodes are combined using the Directed Acyclic Graph (DAG) method to obtain a scene configuration task, which is a task used to construct the target scene model, and there are upstream-downstream relationships among the task nodes in the scene configuration task;

[0082] Obtain the scene configuration data collected from the actual vehicle, wherein the scene configuration data is scene data used to construct the target scene model;

[0083] The scene configuration content is obtained based on the scene configuration data and the scene configuration task.

[0084] In this embodiment, the vehicle simulation platform may further include an automation task module. The automation task module is used to trigger the automatic creation of the target scene model and is responsible for tasks including constructing scene configuration tasks and automatically advancing the execution progress of scene configuration tasks in the cloud.

[0085] Scene configuration content can include scene configuration data and scene configuration tasks. The scene configuration data is collected by sensors on the actual vehicle during its operation, including data collected by sensors such as cameras. The actual vehicle is a real vehicle driven by a driver on a real road. The actual vehicle is equipped with at least one sensor. While the driver is driving the actual vehicle on a real road, the sensors collect scene configuration data of the real road and the surrounding environment. The actual vehicle uploads the collected scene configuration data to the vehicle simulation platform. The vehicle simulation platform creates a target scene model based on the real scene configuration data collected by the actual vehicle, enabling the reconstruction of large-scale world scene models. This solves the problems of insufficient scene diversity and low realism caused by small-batch scene editing and management in related technologies. This application does not impose any restrictions on the driving or operating state of the actual vehicle when collecting scene configuration data. Those skilled in the art can set the system to collect scene configuration data in any state, or they can set it to collect scene configuration data only under certain driving states (such as constant speed driving) according to actual needs.

[0086] Scene configuration tasks can be obtained by combining task nodes. Specifically, a scene configuration task may include one or more tasks for building a target scene model. Each task consists of multiple task nodes, each responsible for a specific function. Task nodes can be flexibly combined, and each task node can be configured with an individual execution mode, such as parallel execution mode, serial execution mode, or a hybrid parallel and serial execution mode. In this embodiment, a Directed Acyclic Graph (DAG) method is used to combine task nodes to obtain the scene configuration task. In an optional embodiment, when designing automated scene configuration tasks, an adjacency list can be used to store the structure of the DAG. During the construction of the DAG, two task nodes with an upstream-downstream relationship need to be input sequentially, and it needs to be confirmed that no cycles appear in the entire structure. Therefore, in the scene configuration task, there is an upstream-downstream relationship between task nodes. The downstream task node will only start when the upstream task node is completed; when there are multiple upstream task nodes, the downstream task node can only start after all upstream task nodes are completed, ensuring the orderliness and continuity of task nodes during execution.

[0087] Reference Figure 2 This is a schematic diagram of the task node combination of an embodiment of a vehicle simulation task processing method of this application.

[0088] Figure 2 This paper illustrates a scenario-based task configuration obtained by combining task nodes using a Directed Acyclic Graph (DAG) method, where upstream and downstream relationships exist between the task nodes. The initial node, marked "Start," serves as the starting point for task scheduling. Following "Start" are three parallel tasks: task node A2011, task node B2012, and task node C2013. These three task nodes, starting from the initial node, can be launched simultaneously. Task node D2021 depends on the completion of task node A2011; task node D2021 can only launch after task node A2011 is completed. Task node E2031 depends on both task nodes B2012 and D2021; task node E2031 can only launch after both tasks B2012 and D2021 are completed. The two parallel tasks, task nodes F2041 and G2042, depend on task node D2021; task nodes F2041 and G2042 can only launch after task node D2021 is completed. Task node H2043 depends on task nodes C2013 and E2031. Task node H2043 can only start after these two task nodes are completed. After all task nodes are completed, specifically task nodes F2041, G2042, and H2043, the process reaches its end and is marked as "complete".

[0089] This application employs a directed acyclic graph (DAG) method to combine task nodes to obtain scene configuration tasks. Because task nodes can be flexibly connected, it offers high flexibility, allowing for easy combination of different task nodes in a "building block" manner. Furthermore, the scene configuration tasks obtained using the DAG method have upstream-downstream relationships between task nodes. In this application embodiment, it can be configured so that when all upstream task nodes complete, the corresponding downstream task node starts execution. This improves the orderliness and continuity of task node execution, significantly enhances the efficiency of task node flow, reduces the need for human intervention, and makes task management more efficient and intelligent.

[0090] In an optional embodiment of this application, the scene configuration task includes a data annotation task and a scene construction task, wherein the data annotation task and the scene construction task each have corresponding task nodes, and step 102 may further include:

[0091] The scene configuration data and the data annotation task are uploaded to the cloud so that the cloud can annotate the scene configuration data;

[0092] Obtain the node status of the task nodes for the data annotation task;

[0093] When the node status is "completed", retrieve the labeled scene configuration data from the cloud.

[0094] Based on the labeled scene configuration data, a task start command is sent to the cloud, which is used to cause the cloud to start executing the scene construction task.

[0095] In this embodiment, the scene configuration task may further include a data annotation task and a scene construction task. By uploading the scene configuration content to the cloud, the cloud is driven by the automated task module to complete the construction of the target scene model.

[0096] First, the scene configuration data and data annotation tasks are uploaded to the cloud. Upon receiving the scene configuration data and data annotation tasks, the cloud annotates the scene configuration data according to the data annotation tasks. Simultaneously, the automated task module of the automotive simulation platform maintains the node status of the task nodes retrieving data annotation tasks from the cloud. When all task nodes of the data annotation tasks are in the "completed" state, the automated task module retrieves the annotated scene configuration data from the cloud. The automated task module then uploads the annotated scene configuration data and the task start command to the cloud, enabling the cloud to execute the scene construction task based on the annotated scene configuration data, thereby obtaining the target scene model constructed in the cloud.

[0097] This application embodiment drives cloud-based task execution through an automated task module and initiates the next task according to the progress of the previous task, thereby automatically completing the construction of the target scene model. This achieves automated scene model construction, making task management more efficient and intelligent.

[0098] In an optional embodiment of this application, based on the above embodiment of obtaining scene configuration content using scene configuration data and scene configuration tasks, and the embodiment of uploading scene configuration content and target simulation tasks to the cloud to complete the construction of the target scene model, the steps of the method when executing the scene configuration task may further include:

[0099] Obtain the current task node and task status of the scene configuration task in the cloud. The current task node is the task node currently being processed in the scene configuration task. There are downstream task nodes of the current task node in the task nodes of the scene configuration task.

[0100] When the task status is "completed", the execution result of the current task node is obtained from the cloud.

[0101] Based on the execution result, a node start command is sent to the cloud, which is used to enable the cloud to start processing the downstream task node.

[0102] In this embodiment, the execution of task nodes in the scenario configuration task is also driven by the automated task module and executed on the cloud. In an optional embodiment, methods for creating nodes, updating node status, and obtaining node results can be set for each task node. For example, the method for creating a node can be to call the interface for creating a scenario configuration task to initialize a new scenario configuration task node; the method for updating the node status can be to call the scenario configuration task status interface to obtain the current scenario configuration task's task node status to determine whether the task node has been completed; the method for obtaining node results can be to output the current task node's result, including success and failure data, which are used by subsequent task nodes.

[0103] During the execution of the scenario configuration task, the automated task module first obtains the structure of the Directed Acyclic Graph (DAG), identifies the current task node, and calls the method to update the node state to refresh the task state of each task node, thereby obtaining the task state of the current task node. The automated task module checks the task state of each task node's upstream task nodes (i.e., the current task node). If all upstream task nodes have completed, it sequentially calls the method to retrieve the node result for each upstream task node, obtains the results from the cloud, and calls the node creation method to send a node start command to the cloud to start the downstream task node. Simultaneously, it passes the execution results of the upstream task nodes to the downstream task nodes. In one embodiment, if the current task node is still in progress, it skips the current task node and waits for the next refresh.

[0104] This application embodiment uses a directed acyclic graph method to combine task nodes to obtain scene configuration tasks, and during execution, the status of task nodes is monitored in real time through an automated task module, and the corresponding task nodes are processed in the cloud, realizing fully automated task flow and providing a complete automated link for the back transmission of labeled scene configuration data and the creation of scene models.

[0105] In an optional embodiment of this application, according to the above embodiment of simulation on an automotive simulation platform, the automotive simulation platform may further include a cache area, and step 103 may further include:

[0106] When the cache does not store the target scene model and / or the target autonomous driving algorithm, the target scene model and / or the target autonomous driving algorithm in the cloud are obtained.

[0107] The method further includes:

[0108] When the cache contains the target scene model and / or the target autonomous driving algorithm, the target scene model and / or the target autonomous driving algorithm in the cache are retrieved.

[0109] To improve the processing efficiency of simulation tasks, multiple target scene models can typically be distributed to multiple target vehicle model objects for parallel simulation. However, cloud deployment solutions for scene models face the following challenges: the scene model assets are too large, and there are too many small files. When a large number of target vehicle model objects concurrently download the target scene model from the cloud, it can cause bandwidth to be completely occupied, triggering download rate limiting, and other problems. Furthermore, the failure to download any file from the scene model assets will render the assets unusable.

[0110] Therefore, in this embodiment, when deploying the target scene model and / or target autonomous driving algorithm for the target vehicle model object, it first checks whether the required data already exists in the cache. If the required target scene model and / or target autonomous driving algorithm data already exists in the cache, the target vehicle model object will directly synchronize the relevant data from the cache to the local machine; if the required target scene model and / or target autonomous driving algorithm data does not exist in the cache, the first target vehicle model object to be deployed will synchronize the relevant data from the cloud.

[0111] In one embodiment, to ensure synchronized distributed consistency, the automotive simulation platform can employ a distributed locking mechanism to ensure that only one target vehicle model object can synchronize data from the cloud at any given time.

[0112] In this embodiment of the application, when deploying the target scene model and / or the target autonomous driving algorithm for the target vehicle model object, existing data is preferentially obtained from the cache, thereby reducing the amount of data downloaded from the cloud and avoiding a large number of target vehicle model objects concurrently downloading the target scene model from the cloud.

[0113] In an optional embodiment of this application, based on the above embodiment of determining whether the cache area stores the target scene model and / or the target autonomous driving algorithm before obtaining it, the vehicle simulation platform may further include a temporary storage area, and the method for obtaining the target scene model and / or the target autonomous driving algorithm from the cloud may further include:

[0114] Store the target scene model and / or the target autonomous driving algorithm in the temporary storage area;

[0115] Verify the target scene model and / or the target autonomous driving algorithm;

[0116] When the target scene model and / or the target autonomous driving algorithm pass the verification, the target scene model and / or the target autonomous driving algorithm are copied to the cache area.

[0117] In this embodiment, if the required target scene model and / or target autonomous driving algorithm data are not present in the cache, the target vehicle model object will synchronize the relevant data from the cloud. During this synchronization, the downloaded target scene model and / or target autonomous driving algorithm are first stored in the temporary storage area of ​​the vehicle simulation platform. In the temporary storage area, the downloaded target scene model and / or target autonomous driving algorithm undergoes integrity verification. After successful verification, it is copied to the cache, and the target vehicle model object then retrieves the target scene model and / or target autonomous driving algorithm from the cache.

[0118] In one alternative embodiment, if the integrity check fails, the next target vehicle model object continues to synchronize the relevant data from the cloud.

[0119] In this embodiment of the application, when the target vehicle model object synchronizes data from the cloud, it does not directly use the data downloaded from the cloud. Instead, the target vehicle model object will only obtain and use the data after the data passes the integrity verification, thereby avoiding the problem of assets becoming unusable due to file download failure in the prior art.

[0120] In one optional embodiment of this application, the target autonomous driving algorithm in the cache is associated with a first usage frequency, and the target scene model in the cache is associated with a second usage frequency. The method further includes:

[0121] When acquiring the target scene model and / or the target autonomous driving algorithm in the cache, update the first usage frequency and / or the second usage frequency;

[0122] Obtain the data storage capacity and storage limit of the cache area;

[0123] When the amount of data stored exceeds the storage limit, data to be deleted is determined from the target autonomous driving algorithm and the target scene model in the cache based on the first usage frequency and the second usage frequency.

[0124] Delete the data to be deleted.

[0125] In this embodiment, each time the target vehicle model object synchronizes the target autonomous driving algorithm and / or target scene model from the cache to the local machine, the first usage frequency and / or second usage frequency are updated, and the data storage volume in the cache is checked to see if it exceeds the storage limit. If the data storage volume in the cache exceeds the storage limit, the vehicle simulation platform will determine the data to be deleted from the target autonomous driving algorithm and target scene model in the cache based on the first and second usage frequencies, and delete the data to be deleted. In one embodiment, the vehicle simulation platform can determine the data corresponding to the lowest usage frequency between the first and second usage frequencies as the data to be deleted.

[0126] This application embodiment removes some target autonomous driving algorithm and target scene model data from the cache based on usage frequency to ensure efficient use of the cache.

[0127] Reference Figure 3 This is a data download diagram of an embodiment of a vehicle simulation task processing method according to this application.

[0128] Figure 3The document illustrates the complete process of obtaining the target autonomous driving algorithm and the target scene model from the target vehicle model object, including the following steps:

[0129] Step S301: Check if the target autonomous driving algorithm and target scene model exist in the cache. If not, proceed to step S302; if they exist, proceed to step S307.

[0130] Step S302: Use a distributed consistency lock mechanism to synchronize the target scene model and / or the target autonomous driving algorithm from the cloud.

[0131] Step S303: Store the downloaded target scene model and / or target autonomous driving algorithm to a temporary storage area.

[0132] Step S304: Perform an integrity check on the downloaded target scene model and / or target autonomous driving algorithm. If the check passes, proceed to step S305; otherwise, proceed to step S306.

[0133] Step S305: Save the downloaded target scene model and / or target autonomous driving algorithm to the cache area.

[0134] Step S306: If the downloaded target scene model and / or target autonomous driving algorithm fail to be verified, the task fails. The next target vehicle model object continues to synchronize the target scene model and / or target autonomous driving algorithm from the cloud.

[0135] Step S307: Synchronize the target autonomous driving algorithm and target scene model in the cache to the local machine.

[0136] Step S308: Update the usage time and frequency of the synchronized target autonomous driving algorithm and target scene model.

[0137] Step S309: Determine whether the data storage volume in the cache exceeds the storage limit. If it does, proceed to step S310; otherwise, complete the acquisition of the target autonomous driving algorithm and the target scene model.

[0138] Step S310: Clean up some data of the target autonomous driving algorithm and target scene model from the cache based on the frequency of use.

[0139] In an optional embodiment of this application, according to the above embodiment of simulation on an automotive simulation platform, after step 105, the method further includes:

[0140] The target autonomous driving algorithm is evaluated based on the target simulation data to obtain evaluation data corresponding to the target autonomous driving algorithm, and the evaluation data represents the performance level of the target autonomous driving algorithm.

[0141] In this embodiment, after the target simulation task is completed, the target autonomous driving algorithm can be evaluated based on the target simulation data to obtain evaluation data corresponding to the target autonomous driving algorithm. This evaluation data reflects the performance level of the target autonomous driving algorithm. In one embodiment, multiple target vehicle model objects can be used to simulate different versions of the target autonomous driving algorithm, collecting target simulation data corresponding to each version. The target simulation data can include simulation data reflecting dimensions such as autonomous driving safety, autonomous driving comfort, and autonomous driving compliance. Based on this target simulation data, different versions of the target autonomous driving algorithm can be evaluated, allowing for a horizontal comparison and assessment of their respective strengths and weaknesses.

[0142] The embodiments of this application can evaluate the target autonomous driving algorithm, thereby obtaining the performance level of the target autonomous driving algorithm, and providing a reference for the subsequent research and development and updating of the target autonomous driving algorithm.

[0143] To enable those skilled in the art to further understand the technical solutions of the embodiments of this application, the following describes... Figure 4 The architecture of the complete technical solution of the embodiments of this application is shown.

[0144] Reference Figure 4 This is a schematic diagram of the automotive simulation platform architecture, representing an embodiment of a vehicle simulation task processing method according to this application.

[0145] The vehicle simulation platform 4200 communicates with both the actual vehicle 4100 and the cloud platform 4300. By analyzing the driving data transmitted from the actual vehicle 4100, problems can be identified, and target simulation tasks can be set accordingly. Scene configuration data transmitted from the actual vehicle 4100 is retrieved and sent to the scene set 4201. The scene set 4201 then sends the scene configuration data to the automated task module 4202 to initiate the construction of the target scene model. The automated task module 4202 triggers the cloud platform 4300 to execute data annotation and scene construction tasks. After the target scene model is constructed on the cloud platform 4300, the scene construction results are stored in the scene set 4201, and simulation tasks are created through the task management module 4203.

[0146] The task management module 4203 triggers task scheduling through the scheduling service 4204. The scheduling service 4204 distributes corresponding target simulation tasks to multiple simulation workers 4205 (i.e., target vehicle model objects, only three are shown in the figure), enabling the simulation workers 4205 to execute the corresponding target simulation tasks under the target scene model, and send the obtained target simulation data to the corresponding evaluation worker 4206 to evaluate the target autonomous driving algorithm corresponding to the target simulation task. The simulation worker 4205 includes a simulation engine 42052, which inputs real vehicle data 42051 and executes the corresponding simulation task using the autonomous driving algorithm 42053 and the scene model 42054. The evaluation worker 4206 includes an automated evaluation module 42062, which evaluates the target autonomous driving algorithm by inputting the simulation output (i.e., target simulation data).

[0147] Afterwards, the evaluation data and target simulation data can be sent back to the task management module 4203 for result display, horizontal and vertical comparison and other operations. At the same time, the evaluation data and target simulation data can be stored in the storage service 4207.

[0148] In this embodiment, the vehicle simulation platform acquires scene configuration content and target simulation tasks; uploads the scene configuration content and target simulation tasks to the cloud, so that the cloud can create a target scene model based on the scene configuration content, and determine at least one target autonomous driving algorithm from autonomous driving algorithms based on the target simulation tasks; acquires the target scene model and target autonomous driving algorithm from the cloud; determines a target vehicle model object from vehicle model objects based on the target simulation tasks; and controls the target vehicle model object to perform simulation using the target autonomous driving algorithm in the target scene model to obtain target simulation data. This embodiment, by storing the autonomous driving algorithm in the cloud and automatically constructing the target scene model in the cloud, builds a stable and reliable cloud deployment environment for the vehicle simulation platform, realizing a complete data backhaul and scene reconstruction link. When a simulation task is triggered, the target scene model and target autonomous driving algorithm can be acquired from the cloud and sent to the vehicle simulation platform, driving the target vehicle model object to execute the simulation of the target autonomous driving algorithm in the target scene model and generating target simulation data, thereby forming a fully automated simulation process for autonomous driving algorithms, effectively adapting to the verification requirements of complex autonomous driving algorithms.

[0149] Compared to simulation platforms in related technologies that cannot provide a large-scale concurrent algorithm verification environment, the automotive simulation platform in this application embodiment can simultaneously select multiple target vehicle models for simulation, thereby supporting large-scale concurrent simulation experiments. It also significantly reduces the cost of relying on physical vehicle-mounted hardware for autonomous driving algorithm simulation in related technologies. Furthermore, this application embodiment constructs a highly realistic target scene model through scene configuration content, improving the mapping accuracy between the simulation environment and real roads, and solving the problems of insufficient scene diversity and low realism in simulation platforms of related technologies.

[0150] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of this application.

[0151] Based on the above embodiments, this embodiment also provides a vehicle simulation task processing device, which is applied in electronic devices such as terminal devices and servers.

[0152] Reference Figure 5 The diagram shows a structural block diagram of an embodiment of a vehicle simulation task processing device according to this application, which may specifically include the following modules:

[0153] Simulation configuration acquisition module 501 is used to acquire scene configuration content and target simulation task;

[0154] The simulation scene creation module 502 is used to upload the scene configuration content and the target simulation task to the cloud, so that the cloud can create a target scene model according to the scene configuration content and determine at least one target autonomous driving algorithm from the autonomous driving algorithms according to the target simulation task;

[0155] The cloud data acquisition module 503 is used to acquire the target scene model and the target autonomous driving algorithm in the cloud.

[0156] The simulation object acquisition module 504 is used to determine the target vehicle model object from the vehicle model objects according to the target simulation task;

[0157] The simulation task execution module 505 is used to control the target vehicle model object to perform simulation in the target scene model using the target autonomous driving algorithm to obtain target simulation data.

[0158] Optionally, the scene configuration content includes scene configuration data and scene configuration tasks; the vehicle simulation platform includes task nodes; the vehicle simulation platform is communicatively connected to the actual vehicle; and the simulation configuration acquisition module 501 includes:

[0159] The task node combination submodule is used to combine at least two task nodes using a directed acyclic graph method to obtain a scene configuration task. The scene configuration task is a task used to construct the target scene model. There is an upstream-downstream relationship between the task nodes in the scene configuration task.

[0160] The configuration data acquisition submodule is used to acquire the scene configuration data collected by the actual vehicle, wherein the scene configuration data is scene data used to construct the target scene model;

[0161] The configuration content determination submodule is used to obtain the scene configuration content based on the scene configuration data and the scene configuration task.

[0162] Optionally, the scene configuration task includes a data annotation task and a scene construction task, wherein the data annotation task and the scene construction task each have corresponding task nodes, and the simulation scene creation module 502 includes:

[0163] The data annotation submodule is used to upload the scene configuration data and the data annotation task to the cloud, so that the cloud can annotate the scene configuration data;

[0164] The node status acquisition submodule is used to acquire the node status of the task nodes of the data annotation task;

[0165] The annotation data acquisition submodule is used to acquire the annotated scene configuration data from the cloud when the node status is "completed".

[0166] The scene construction submodule is used to send a task start command to the cloud based on the labeled scene configuration data. The task start command is used to cause the cloud to start executing the scene construction task.

[0167] Optionally, the device further includes:

[0168] The task status acquisition module is used to acquire the current task node and the task status of the scene configuration task in the cloud. The current task node is the task node currently being processed in the scene configuration task. There are downstream task nodes of the current task node in the task nodes of the scene configuration task.

[0169] The execution result acquisition module is used to acquire the execution result of the current task node from the cloud when the task status is completed;

[0170] The downstream node startup module is used to send a node startup command to the cloud based on the execution result. The node startup command is used to cause the cloud to start processing the downstream task node.

[0171] Optionally, the vehicle simulation platform includes a cache area, and the cloud data acquisition module 503 includes:

[0172] The first data acquisition submodule is used to acquire the target scene model and / or the target autonomous driving algorithm in the cloud when the cache does not store the target scene model and / or the target autonomous driving algorithm.

[0173] The device further includes:

[0174] The second data acquisition module is used to acquire the target scene model and / or the target autonomous driving algorithm in the cache when the cache stores the target scene model and / or the target autonomous driving algorithm.

[0175] Optionally, the vehicle simulation platform includes a temporary storage area, and the cloud data acquisition module 503 further includes:

[0176] A temporary storage submodule is used to store the target scene model and / or the target autonomous driving algorithm to the temporary storage area;

[0177] The data verification submodule is used to verify the target scene model and / or the target autonomous driving algorithm;

[0178] The data copying submodule is used to copy the target scene model and / or the target autonomous driving algorithm to the cache area when the target scene model and / or the target autonomous driving algorithm passes the verification.

[0179] Optionally, the target autonomous driving algorithm in the cache is associated with a first usage frequency, and the target scene model in the cache is associated with a second usage frequency. The device further includes:

[0180] The frequency update module is used to update the first usage frequency and / or the second usage frequency when acquiring the target scene model and / or the target autonomous driving algorithm in the cache area;

[0181] The cache metrics acquisition module is used to acquire the data storage volume and storage limit of the cache area;

[0182] The data to be deleted determination module is used to determine data to be deleted from the target autonomous driving algorithm and the target scene model in the cache area based on the first usage frequency and the second usage frequency when the data storage volume is greater than the limit storage volume;

[0183] The data deletion module is used to delete the data to be deleted.

[0184] Optionally, the device further includes:

[0185] The algorithm evaluation module is used to evaluate the target autonomous driving algorithm based on the target simulation data, and obtain the evaluation data corresponding to the target autonomous driving algorithm. The evaluation data represents the performance level of the target autonomous driving algorithm.

[0186] This application also provides a computer program product, including a computer program / instructions, which, when executed by a processor, can implement the vehicle simulation task processing method as described in this application embodiment.

[0187] This application also provides a non-volatile readable storage medium storing one or more modules (programs). When these modules are applied to a device, they enable the device to execute the instructions for the method steps in this application.

[0188] This application provides one or more machine-readable media storing instructions that, when executed by one or more processors, cause an electronic device to perform one or more of the methods described in the above embodiments. In this application, the electronic device includes various types of devices such as terminal devices and servers (clusters).

[0189] The embodiments of this disclosure can be implemented as an apparatus configured as desired using any suitable hardware, firmware, software, or any combination thereof, including electronic devices such as terminal devices, servers (clusters), etc. Figure 6 An exemplary apparatus 600 is schematically shown that can be used to implement the various embodiments described in this application.

[0190] In one embodiment, Figure 6 An exemplary device 600 is shown, which includes one or more processors 602, a control module (chipset) 604 coupled to at least one of the processors 602, a memory 606 coupled to the control module 604, a non-volatile memory (NVM) / storage device 608 coupled to the control module 604, one or more input / output devices 610 coupled to the control module 604, and a network interface 612 coupled to the control module 604.

[0191] Processor 602 may include one or more single-core or multi-core processors, and processor 602 may include any combination of general-purpose processors or special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). In some embodiments, device 600 can serve as a terminal device, server (cluster), or other device as described in the embodiments of this application.

[0192] In some embodiments, the apparatus 600 may include one or more computer-readable media (e.g., memory 606 or NVM / storage device 608) having instructions 614 and one or more processors 602 that are combined with the one or more computer-readable media and configured to execute the instructions 614 to implement the module and thus perform the actions described in this disclosure.

[0193] In one embodiment, the control module 604 may include any suitable interface controller to provide any suitable interface to at least one of the processors 602 and / or any suitable device or component communicating with the control module 604.

[0194] The control module 604 may include a memory controller module to provide an interface to the memory 606. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0195] Memory 606 may be used, for example, to load and store data and / or instructions 614 for device 600. In one embodiment, memory 606 may include any suitable volatile memory, such as suitable DRAM. In some embodiments, memory 606 may include double data rate type quad synchronous dynamic random access memory (DDR4 SDRAM).

[0196] In one embodiment, the control module 604 may include one or more input / output controllers to provide an interface to the NVM / storage device 608 and (one or more) input / output devices 610.

[0197] For example, NVM / storage device 608 may be used to store data and / or instructions 614. NVM / storage device 608 may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (one or more) non-volatile storage devices (e.g., one or more hard disk drives (HDDs), one or more optical disc drives (CDs), and / or one or more digital universal optical disc (DVD) drives).

[0198] NVM / storage device 608 may include storage resources that are physically part of a device on which device 600 is mounted, or that are accessible to the device but do not necessarily have to be part of the device. For example, NVM / storage device 608 may be accessed via a network through one or more input / output devices 610.

[0199] One or more input / output devices 610 may provide an interface for device 600 to communicate with any other suitable device. Input / output devices 610 may include communication components, audio components, sensor components, etc. A network interface 612 may provide an interface for device 600 to communicate via one or more networks. Device 600 may wirelessly communicate with one or more components of a wireless network according to any of one or more wireless network standards and / or protocols, such as accessing wireless networks based on communication standards, such as WiFi, 2G, 3G, 4G, 5G, etc., or combinations thereof.

[0200] In one embodiment, at least one of the processors 602 may be logically packaged with one or more controllers (e.g., memory controller modules) of the control module 604. In one embodiment, at least one of the processors 602 may be logically packaged with one or more controllers of the control module 604 to form a system-in-package (SiP). In one embodiment, at least one of the processors 602 may be integrated with the logic of one or more controllers of the control module 604 on the same die. In one embodiment, at least one of the processors 602 may be integrated with the logic of one or more controllers of the control module 604 on the same die to form a system-on-a-chip (SoC).

[0201] In various embodiments, device 600 may be, but is not limited to, a server, desktop computing device, or mobile computing device (e.g., laptop, handheld computing device, tablet, netbook, etc.). In various embodiments, device 600 may have more or fewer components and / or different architectures. For example, in some embodiments, device 600 includes one or more cameras, a keyboard, a liquid crystal display (LCD) screen (including a touchscreen display), a non-volatile memory port, multiple antennas, a graphics chip, an application-specific integrated circuit (ASIC), and a speaker.

[0202] The detection device can use a main control chip as a processor or control module, and sensor data, position information, etc. can be stored in a memory or NVM / storage device. The sensor group can be used as an input / output device, and the communication interface can include a network interface.

[0203] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0204] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0205] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other processing terminal device for a programmable vehicle simulation task to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable vehicle simulation task processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0206] These computer program instructions may also be stored in a computer-readable storage medium that can direct a processing terminal device, such as a computer or other programmable vehicle simulation task, to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0207] These computer program instructions can also be loaded onto a computer or other programmable vehicle simulation task processing terminal device, causing a series of operational steps to be executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0208] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0209] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0210] The present application provides a detailed description of a vehicle simulation task processing method and apparatus, an electronic device, and a storage medium. Specific examples have been used to illustrate the principles and implementation methods of the present application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present application. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of the present application. Therefore, the content of this specification should not be construed as a limitation of the present application.

Claims

1. A method for processing vehicle simulation tasks, characterized in that, The method, applied to an automotive simulation platform including vehicle model objects and connected to a cloud computing platform storing autonomous driving algorithms, includes: Obtain the scene configuration content and target simulation task; The scenario configuration content and the target simulation task are uploaded to the cloud so that the cloud can create a target scenario model based on the scenario configuration content and determine at least one target autonomous driving algorithm from the autonomous driving algorithms based on the target simulation task. Obtain the target scene model and the target autonomous driving algorithm from the cloud; Based on the target simulation task, determine the target vehicle model object from the vehicle model object; The target vehicle model object is controlled to perform simulation using the target autonomous driving algorithm in the target scene model to obtain target simulation data.

2. The method according to claim 1, characterized in that, The scene configuration content includes scene configuration data and scene configuration tasks. The vehicle simulation platform includes task nodes and is communicatively connected to the actual vehicle. Obtaining the scene configuration content includes: At least two of the task nodes are combined using a directed acyclic graph method to obtain a scene configuration task, which is a task used to construct the target scene model, and there are upstream-downstream relationships among the task nodes in the scene configuration task; Obtain the scene configuration data collected from the actual vehicle, wherein the scene configuration data is scene data used to construct the target scene model; The scene configuration content is obtained based on the scene configuration data and the scene configuration task.

3. The method according to claim 2, characterized in that, The scene configuration task includes a data annotation task and a scene construction task. The data annotation task and the scene construction task each have corresponding task nodes. Uploading the scene configuration content and the target simulation task to the cloud includes: The scene configuration data and the data annotation task are uploaded to the cloud so that the cloud can annotate the scene configuration data; Obtain the node status of the task nodes for the data annotation task; When the node status is "completed", retrieve the labeled scene configuration data from the cloud. Based on the labeled scene configuration data, a task start command is sent to the cloud, which is used to cause the cloud to start executing the scene construction task.

4. The method according to any one of claims 2-3, characterized in that, The method further includes: Obtain the current task node and task status of the scene configuration task in the cloud. The current task node is the task node currently being processed in the scene configuration task. There are downstream task nodes of the current task node in the task nodes of the scene configuration task. When the task status is "completed", the execution result of the current task node is obtained from the cloud. Based on the execution result, a node start command is sent to the cloud, which is used to enable the cloud to start processing the downstream task node.

5. The method according to any one of claims 1-4, characterized in that, The vehicle simulation platform includes a cache area, and the step of acquiring the target scene model and the target autonomous driving algorithm from the cloud includes: When the cache does not store the target scene model and / or the target autonomous driving algorithm, the target scene model and / or the target autonomous driving algorithm in the cloud are obtained. The method further includes: When the cache contains the target scene model and / or the target autonomous driving algorithm, the target scene model and / or the target autonomous driving algorithm in the cache are retrieved.

6. The method according to claim 5, characterized in that, The vehicle simulation platform includes a temporary storage area. After acquiring the target scene model and / or the target autonomous driving algorithm from the cloud, the method includes: Store the target scene model and / or the target autonomous driving algorithm in the temporary storage area; Verify the target scene model and / or the target autonomous driving algorithm; When the target scene model and / or the target autonomous driving algorithm pass the verification, the target scene model and / or the target autonomous driving algorithm are copied to the cache area.

7. The method according to claim 5 or 6, characterized in that, The target autonomous driving algorithm in the cache is associated with a first usage frequency, and the target scene model in the cache is associated with a second usage frequency. The method further includes: When acquiring the target scene model and / or the target autonomous driving algorithm in the cache, update the first usage frequency and / or the second usage frequency; Obtain the data storage capacity and storage limit of the cache area; When the amount of data stored exceeds the storage limit, data to be deleted is determined from the target autonomous driving algorithm and the target scene model in the cache based on the first usage frequency and the second usage frequency. Delete the data to be deleted.

8. The method according to any one of claims 1-7, characterized in that, After controlling the target vehicle model object to perform simulation using the target autonomous driving algorithm in the target scene model to obtain target simulation data, the method further includes: The target autonomous driving algorithm is evaluated based on the target simulation data to obtain evaluation data corresponding to the target autonomous driving algorithm, and the evaluation data represents the performance level of the target autonomous driving algorithm.

9. A processing device for vehicle simulation tasks, characterized in that, The device is applied to an automotive simulation platform, which includes vehicle model objects and is connected to a cloud for communication. The cloud stores autonomous driving algorithms. The simulation configuration acquisition module is used to acquire scene configuration content and target simulation tasks; The simulation scene creation module is used to upload the scene configuration content and the target simulation task to the cloud, so that the cloud can create a target scene model according to the scene configuration content and determine at least one target autonomous driving algorithm from the autonomous driving algorithms according to the target simulation task; The cloud data acquisition module is used to acquire the target scene model and the target autonomous driving algorithm in the cloud. The simulation object acquisition module is used to determine the target vehicle model object from the vehicle model objects according to the target simulation task; The simulation task execution module is used to control the target vehicle model object to perform simulation in the target scene model using the target autonomous driving algorithm to obtain target simulation data.

10. An electronic device, characterized in that, include: processor; and A memory having executable code stored thereon, which, when executed, causes the processor to perform the processing method for the vehicle simulation task as described in any one of claims 1-8.

11. One or more machine-readable media having executable code stored thereon, which, when executed, causes a processor to perform a processing method for a vehicle simulation task as described in any one of claims 1-8.

12. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the processing method for the vehicle simulation task as described in any one of claims 1-8.

Images