Autonomous driving training methods, devices, systems, equipment, and storage media

Abstract

The present disclosure provides an autonomous driving training method, apparatus, system, device, storage medium, and computer program product, relating to the fields of artificial intelligence technology, particularly autonomous driving and cloud platform technology. The specific implementation scheme comprises: receiving an autonomous driving training request and determining the training task specified in the autonomous driving training request; obtaining a basic training image, as well as a training dataset and task code package corresponding to the training task; adding the training dataset and task code package to the basic training image to obtain a training image; creating a training container based on the training image, and rehearsing the training task through the training container. This improves the efficiency of autonomous driving training.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, specifically to the field of autonomous driving and cloud platform technology, and more particularly to an autonomous driving training method, apparatus, system, equipment, storage medium, and computer program product. Background Art

[0002] With the increasing adoption of artificial intelligence and fifth-generation mobile communications, the use of autonomous driving technology in vehicles is increasing. Consequently, there is an urgent need to train a large number of professionals to learn and develop autonomous driving-related technologies to adapt to the industry's rapid development. Because autonomous driving technology involves the intersection and integration of multidisciplinary knowledge, the content is fragmented and complex, equipping students with relevant knowledge requires practical training. Summary of the Invention

[0003] The present disclosure provides an autonomous driving training method, apparatus, system, device, storage medium, and computer program product, which improve the efficiency of autonomous driving training.

[0004] According to one aspect of the present disclosure, a method for autonomous driving training is provided, including: receiving an autonomous driving training request and determining a training task specified by the autonomous driving training request; obtaining a basic training image, and a training data set and a task code package corresponding to the training task; adding the training data set and the task code package to the basic training image to obtain an image to be trained; creating a training container based on the image to be trained, and rehearsing the training task through the training container.

[0005] According to another aspect of the present disclosure, an autonomous driving training device is provided, including: a receiving module, configured to receive an autonomous driving training request and determine a training task specified by the autonomous driving training request; an acquisition module, configured to obtain a basic training image, and a training data set and task code package corresponding to the training task; an adding module, configured to add the training data set and task code package to the basic training image to obtain an image to be trained; and a rehearsal module, configured to create a training container based on the image to be trained, and rehearse the training task through the training container.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the autonomous driving training method.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the autonomous driving training method.

[0008] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, which implements the above-mentioned autonomous driving training method when executed by a processor.

[0009] According to another aspect of the present disclosure, an autonomous driving training platform is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the autonomous driving training method.

[0010] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0012] Figure 1 is an exemplary system architecture diagram in which the present disclosure may be applied;

[0013] Figure 2 is a flowchart of an embodiment of the autonomous driving training method according to the present disclosure;

[0014] Figure 3 is a flowchart of another embodiment of the autonomous driving training method according to the present disclosure;

[0015] Figure 4 is a flowchart of another embodiment of the autonomous driving training method according to the present disclosure;

[0016] Figure 5 is a structural diagram of an embodiment of an autonomous driving training device according to the present disclosure;

[0017] Figure 6 is a structural diagram of an embodiment of an autonomous driving training system according to the present disclosure;

[0018] Figure 7 is a schematic diagram of the resource architecture of the cloud platform in the autonomous driving training system according to the present disclosure;

[0019] Figure 8 This is a schematic diagram of the application process of the autonomous driving training method disclosed in the present invention on a cloud platform;

[0020] Figure 9 2 is a block diagram of an electronic device used to implement the autonomous driving training method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0021] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0022] Figure 1 An exemplary system architecture 100 is shown to which an embodiment of the autonomous driving training method disclosed herein may be applied.

[0023] like Figure 1 As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Network 102 is used to provide a medium for a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0024] The terminal device 101 can interact with the server 103 through the network 102. A web browser application or a training service application can be installed on the terminal device 101. The server 103 can provide various data services. For example, a user can log in to the server 103 through the web browser on the terminal device 101 and initiate a training request. The server 103 can provide the user with relevant training services based on the training request.

[0025] It should be noted that the terminal device 101 can be hardware or software. When the terminal device 101 is hardware, it can be various electronic devices, including but not limited to smart phones, tablet computers, laptop computers and desktop computers, etc. When the terminal device 101 is software, it can be installed in the above-mentioned electronic devices. It can be implemented as multiple software or software modules, or as a single software or software module. No specific limitation is made here. The server 103 can be hardware or software. When the server 103 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server 103 is software, it can be implemented as multiple software or software modules (for example, for providing distributed services), or as a single software or software module. No specific limitation is made here.

[0026] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0027] It should be noted that the autonomous driving training method provided in the embodiments of the present disclosure is generally executed by the server 103 , and accordingly, the autonomous driving training device is generally set in the server 103 .

[0028] Continue to refer Figure 2 , which shows a process 200 of an embodiment of the autonomous driving training method according to the present disclosure. The method includes the following steps:

[0029] Step 201: Receive an autonomous driving training request and determine the training task specified in the autonomous driving training request.

[0030] In this embodiment, the execution subject of the autonomous driving training method (e.g. Figure 1 The server 103 shown in FIG. 103 may first obtain an autonomous driving training request, wherein the autonomous driving training request may be a request received by a learner through a terminal (e.g., Figure 1 The autonomous driving training in this embodiment is mainly used to help learners understand the software functions and architecture of autonomous driving, so that learners can perform testing, operation and maintenance, and secondary development. After obtaining the autonomous driving training request, the above-mentioned execution entity can further determine the training tasks specified by the autonomous driving training request. In some optional implementations of this embodiment, when initiating a training request, the learner can independently specify the training tasks they want to complete. In other optional implementations of this embodiment, the above-mentioned execution entity can select training tasks for the learner based on the learner's previous training records and training results. For example, if the learner's previous training records are not obtained, it means that this is his first time to participate in the training, and a basic training task can be selected for him; if the learner's previous training records are obtained, the training tasks with unqualified results in the previous training can be selected for this training.

[0031] In some optional implementations of this embodiment, the autonomous driving training task includes at least one of the following: positioning training task, high-precision map training task, perception training task, decision-making and planning training task, control training task and middleware communication training task. In this embodiment, positioning refers to determining the position of the vehicle, high-precision map refers to constructing a high-precision map, perception refers to environmental perception through the perception data of various sensors on the vehicle, decision-making and planning refers to path planning and behavior decision-making of the autonomous driving vehicle, control refers to controlling the behavior of the autonomous driving vehicle, and middleware communication mainly refers to the communication function of the vehicle network. Accordingly, each function corresponds to a training task, which can be used to learn and debug the data and code related to the autonomous driving function. For example, the perception training task can debug and analyze the perception data packets of various sensors of the autonomous driving vehicle.

[0032] Step 202: Obtain a basic training image, as well as a training data set and task code package corresponding to the training task.

[0033] In this embodiment, after determining the training task, the executor of the autonomous driving training method can obtain the training data set and task code package required for the training task, and can also obtain the basic training image. Specifically, in different training tasks, the training data sets required are also different. For example, the training data set used for the perception training task is mainly the perception data of the on-board sensors, including image data collected by the image sensor and / or point cloud data collected by the lidar; while the training data set used for the control training task can be a real road test data. Similarly, the code to implement each training function is also different, and each training task has its own corresponding task code package.

[0034] It should be noted that both the training data set and the task code package are pre-saved in the storage space of the above-mentioned execution entity and are bound to a specific training task. In addition, the basic training image is also saved in the aforementioned storage space. A mirror is a form of file storage. The data on one disk has an identical copy on another disk, which is called a mirror. The basic training image in this embodiment can be considered as a template that includes various environments and services related to autonomous driving, which is used to provide other basic functions in addition to the training tasks in the subsequent training process.

[0035] Step 203: Add the training data set and the task code package to the basic training image to obtain the image to be trained.

[0036] In this embodiment, after obtaining the basic training image, training data set and task code package, the above-mentioned execution entity can add the training data set and task code package into the basic training image. Specifically, the training data set can be used as the data to be processed in subsequent training, and the task code package can be used as the functional code for processing the training data set. After adding it to the basic training image, the image to be trained is obtained.

[0037] It should be noted that since the subsequent training mainly focuses on learning and debugging the task codes corresponding to the training tasks, the task code package in this embodiment exists in the form of source code to facilitate students' learning, and the basic training image exists in the form of binary code, as long as the relevant basic functions are implemented.

[0038] Step 204: Create a training container based on the image to be trained, and practice the training task through the training container.

[0039] In this embodiment, after obtaining the image to be trained, the execution entity can package the image into a standardized unit, namely a container. A container is an instance of an image, and its status can include running, stopped, deleted, and paused. The container's operation is the process of practicing the training task using the training container. Trainees complete practical training in the corresponding autonomous driving function by observing and learning from the entire practice process.

[0040] The autonomous driving training method provided by the disclosed embodiment first determines the training task through an autonomous driving training request, then obtains a basic training image, as well as a training data set and task code package corresponding to the training task. The training data set and task code package are then added to the basic training image to obtain the image to be trained. Finally, a training container is created based on the image to be trained, and the training task is rehearsed through the training container. Using the pre-saved training data set and task code package, a training image can be quickly generated, and a training container can be created to complete the rehearsal of the training task. In this process, there is no need for complex system installation and environment deployment, thereby significantly improving the efficiency of autonomous driving training.

[0041] Further reference Figure 3 , which shows a process 300 of another embodiment of the autonomous driving training method according to the present disclosure. The method includes the following steps:

[0042] Step 301: Receive an autonomous driving training request and determine the training task specified in the autonomous driving training request.

[0043] In this embodiment, the specific operation of step 301 has been Figure 2 Step 201 is described in detail in the illustrated embodiment and will not be repeated here.

[0044] Step 302: Obtain a basic training image, and specify multiple training tasks in response to an autonomous driving training request, and obtain a training data set and task code package corresponding to each training task.

[0045] In this embodiment, after determining the training tasks specified in the autonomous driving training request, the executing entity of the autonomous driving training method can first determine the number of training tasks. If it is determined that the autonomous driving training request specifies multiple training tasks, it can obtain the training dataset and task code package corresponding to each training task, and simultaneously obtain the basic training image. The training dataset, task code package, and basic training image are all pre-stored in a specific storage space, and the executing entity can directly access this storage space and obtain the required data from it.

[0046] Step 303: Rearrange the obtained multiple training data sets and multiple task code packages to obtain target training data.

[0047] In this embodiment, after obtaining the training data set and task code package corresponding to each training task, the above-mentioned execution entity can extract the contents of the multiple training data sets and multiple task code packages obtained, and then rearrange them according to a predetermined format to obtain the target training data. For example, if the obtained task code package includes both a positioning training task code package and a control training task code package, the source code contained in the above two task code packages can be rearranged according to the format of the code package so that the two can cooperate with each other in terms of function. At the same time, if there is a correlation between the data in the multiple training data sets, they can be associated and / or fused; if there is no correlation, all of them can be directly saved. Ultimately, the target training data can be obtained.

[0048] Step 304: Add the target training data to the basic training image to obtain the image to be trained.

[0049] In this embodiment, the execution entity can directly add the acquired target training data to the basic training image to obtain the image to be trained. The data in the target training data that originates from the aforementioned multiple training data sets can be used as the data to be processed in subsequent training, while the source code from the aforementioned multiple task code packages can be used as the functional code for the training task.

[0050] Step 305: Obtain training configuration information.

[0051] In this embodiment, after obtaining the image to be trained, the executor of the autonomous driving training method can further obtain training configuration information. This training configuration information may include information such as the name, deployment version, container port, and available memory configured by the current trainee for the container. It may also include the training task name specified for this training and the corresponding training dataset name and version.

[0052] Step 306: Generate container deployment information based on the training configuration information.

[0053] In this embodiment, the executor of the autonomous driving training method, after obtaining the training configuration information, can generate container deployment information based on the training configuration information. During the containerized deployment process, each step in the deployment operation can be implied and then concentrated into a script to complete the original complex deployment process. The container deployment information generated in this embodiment can also be a deployment file, which mainly determines the computing resources, container resources and other resource information required for the subsequent training process based on the training configuration information.

[0054] Step 307: Create a training container based on the container deployment information and the image to be trained, and practice the training task through the training container.

[0055] In this embodiment, the execution entity can first configure computing resources and container resources based on the container deployment information, then execute the deployment operation commands stored in the container deployment information one by one, thereby creating and starting one or more training containers from the image to be trained. The training task can then be practiced by running the training containers.

[0056] In some optional implementations of this embodiment, if a training request specifies multiple training tasks, after obtaining multiple training data sets and multiple task code packages, each training data set and the corresponding task code package can be added to the basic training image. In this case, multiple training task images will be obtained. Afterwards, in the process of obtaining the training configuration information, the above-mentioned multiple training task images can be fused. Specifically, based on the storage location of the training data set and the task code package, the training data set file and the task code package file can be obtained. Then, the data content of these files can be extracted and rearranged in a certain format. The fused task image can be obtained as the image to be trained.

[0057] In this embodiment, the executor of the autonomous driving training method can rearrange multiple training data sets and multiple task code packages corresponding to multiple training tasks, and then add the obtained target training data to the basic training image to generate an image to be trained. In the scenario of multiple training tasks, it is no longer necessary to generate a training image for each task separately. Instead, a single image that can complete multiple tasks at the same time is generated by code rearrangement, which greatly improves the efficiency of image generation, thereby improving the training efficiency of multi-task training. At the same time, this embodiment also packages the code and operating environment required for the autonomous driving training function in the training container through containerized deployment, thereby realizing unified deployment of the environment and tasks. When rehearsing training tasks, there is no need to deploy the development environment separately, which improves the portability of training tasks.

[0058] Further reference Figure 4 , which shows a process 400 of another embodiment of the autonomous driving training method according to the present disclosure. The processing method includes the following steps:

[0059] Step 401: Receive an autonomous driving training request and determine the training task specified in the autonomous driving training request.

[0060] Step 402: Obtain a basic training image, as well as a training data set and task code package corresponding to the training task.

[0061] In this embodiment, the specific operations of steps 401-402 are already in Figure 2 In the illustrated embodiment, steps 201 - 202 are described in detail and will not be repeated here.

[0062] Step 403: Receive a modification operation on the task code package and obtain a user code package.

[0063] In this embodiment, the execution subject of the autonomous driving training method can also receive modification operations on the task code package, and modify the source code in the task code package according to the modification operation to obtain a user code package. It is understandable that the modification operation here can be initiated by the trainee currently undergoing training. The trainee can make appropriate modifications to the code package based on what he has learned, and determine the autonomous driving results caused by the current modification through subsequent practice steps. Specifically, the current execution subject can provide an online code editing service to the terminal, for example, through an online code editor, to receive the modification operation of the task code package by the terminal user (trainee).

[0064] It is understood that during the training process, trainees can not only study and observe the original code, but also modify the original code according to their own understanding to obtain user code, and further determine their learning effect based on the training effect of the user code. This embodiment allows trainees to make personalized changes to the original code by receiving modification operations on the task code package, facilitating the evaluation of their personal training results and improving the flexibility and comprehensiveness of the training.

[0065] Step 404: Add the training data set and the user code package to the basic training image to obtain the image to be trained.

[0066] In this embodiment, step 404 and Figure 2 In the illustrated embodiment, step 203 is substantially the same, with the only difference being that the task code package is replaced with a user code package. The specific operation has been described in detail in step 203 and will not be repeated here.

[0067] Step 405: Create a training container based on the image to be trained.

[0068] In this embodiment, the method of creating a training container in step 405 is the same as Figure 2 In the illustrated embodiment, the method for creating a training container in step 204 is the same, and the specific operation has been described in detail in step 204 and will not be repeated here.

[0069] In this embodiment, there are two ways to practice the training task using the training container, one of which is described in steps 406-408 below, and the other is described in steps 409-410 below. It should be noted that the autonomous driving training method of this embodiment can include both steps 406-410, or only one of 406-408 or 409-410.

[0070] Step 406: Practice the training task step by step using the training container according to the training steps.

[0071] In this embodiment, the entire training task can be divided into multiple executable training steps. When practicing the training task through the training container, the training can be carried out step by step according to the training steps. For example, when practicing a perception training task, if the data in the perception training dataset is image data collected by an image sensor, the training steps can be divided into steps such as removing image noise, extracting the target object, and identifying the target object.

[0072] Those skilled in the art should understand that the above-mentioned step examples are merely illustrative examples. In actual applications, the training steps can be specifically divided according to the goals and requirements of different training tasks, and the embodiments of the present application do not limit this.

[0073] Step 407: In response to the completion of a drill step, an output result corresponding to the drill step is obtained.

[0074] In this embodiment, after completing a training step, the subject executing the autonomous driving training method can further obtain the training results output after the training step. Similarly, taking the training step of the perception training task in step 406 as an example, after completing the image noise removal training step, the denoised image can be obtained as the output result; and after completing the target object recognition training step, the recognition result can be obtained as the output result.

[0075] Step 408: Evaluate the output result and output the evaluation result.

[0076] In this embodiment, the execution entity may further evaluate the output result and output the evaluation result. Specifically, the output result may be compared with a pre-stored standard result. If the output result meets the standard result requirements, a passing result is output; if the output result does not meet the standard result requirements, a failing result is output. The standard result requirements may be a unique result, such as a fixed code or a specific recognition result (such as a pedestrian); or a category of results, such as code data that meets specific format requirements.

[0077] This embodiment implements point-by-point evaluation by rehearsing the training tasks step by step and outputting the evaluation results of the rehearsal steps in real time. It can timely determine the learning effect of the trainees during the training process, and improve the pertinence and timeliness of the training effect evaluation.

[0078] Step 409: Under different driving scenarios, the training task is simulated multiple times through the training container to obtain multiple simulation results.

[0079] In this embodiment, the execution entity simulates the training task multiple times in different driving scenarios using the training container. For example, if the current training task targets a traffic light scenario, data for multiple different intersection scenarios can be obtained, including straight-ahead or turning scenarios at an intersection, straight-ahead scenarios at a T-junction, and so on. This scenario data is then added to the training container, allowing the training container to simulate multiple driving scenarios separately and obtain the simulation results for each.

[0080] Step 410: Evaluate multiple simulation results in parallel and output the evaluation results.

[0081] After obtaining multiple simulation results, the execution entity can evaluate each simulation result separately. If the simulation results in each driving scenario meet the application requirements, the simulation passing evaluation result is output; if the simulation results in any driving scenario do not meet the application requirements, the simulation failing evaluation result is output, thereby achieving parallel evaluation of multiple simulation results. Specific evaluation criteria may include result accuracy evaluation and / or driving safety evaluation, wherein the result accuracy evaluation can be used to determine whether the autonomous driving vehicle is driving according to the set rules (e.g., a set route), and the driving safety evaluation can be used to determine whether the autonomous driving vehicle meets traffic safety requirements during driving.

[0082] This embodiment can objectively and comprehensively evaluate the training effect by performing parallel evaluation on the simulation results of the training task in multiple driving scenarios.

[0083] Further references Figure 5 , as a response to the above Figure 2-Figure 4 The present disclosure provides an embodiment of an automatic driving training device. Figure 2-Figure 4 Corresponding to the method embodiment shown, the device can be specifically applied in a server.

[0084] like Figure 5 As shown, the autonomous driving training device 500 of this embodiment may include a receiving module 501, an acquisition module 502, an addition module 503, and a rehearsal module 504. The receiving module 501 is configured to receive an autonomous driving training request and determine the training task specified in the autonomous driving training request; the acquisition module 502 is configured to obtain a basic training image, as well as a training dataset and task code package corresponding to the training task; the addition module 503 is configured to add the training dataset and task code package to the basic training image to obtain an image to be trained; and the rehearsal module 504 is configured to create a training container based on the image to be trained and rehearse the training task using the training container.

[0085] In this embodiment, in the automatic driving training device 500, the specific processing of the receiving module 501, the obtaining module 502, the adding module 503 and the rehearsal module 504 and the technical effects thereof can be referred to respectively. Figure 2 The relevant descriptions of steps 201-204 in the corresponding embodiment are not repeated here.

[0086] In some optional implementations of this embodiment, the autonomous driving training task includes at least one of the following: positioning training task, high-precision map training task, perception training task, decision-making and planning training task, control training task and middleware communication training task.

[0087] In some optional implementations of this embodiment, the acquisition module 502 includes: a first acquisition unit, configured to obtain a basic training image, and to specify multiple training tasks in response to an autonomous driving training request, and to obtain a training data set and task code package corresponding to each training task respectively.

[0088] The adding module 503 includes: an arrangement unit, configured to rearrange the obtained multiple training data sets and multiple task code packages to obtain target training data; a first adding unit, configured to add the target training data into the basic training image to obtain the image to be trained.

[0089] In some optional implementations of this embodiment, the rehearsal module 504 includes: a second acquisition unit, configured to acquire training configuration information; a generation unit, configured to generate container deployment information based on the training configuration information; and a creation unit, configured to create a training container based on the container deployment information and the image to be trained.

[0090] In some optional implementations of this embodiment, the autonomous driving training device 500 further includes: a modification module configured to receive a modification operation on the task code package and obtain a user code package.

[0091] The adding module 503 includes: a second adding unit configured to add the training data set and the user code package into the basic training image to obtain the image to be trained.

[0092] In some optional implementations of this embodiment, the rehearsal module 504 includes: a rehearsal unit, configured to rehearse the training task step by step through the training container according to the rehearsal steps; a result acquisition unit, configured to obtain the output result corresponding to a rehearsal step in response to the completion of the execution of the rehearsal step; and a first evaluation unit, configured to evaluate the output result and output the evaluation result.

[0093] In some optional implementations of this embodiment, the rehearsal module 504 includes: a simulation unit, configured to simulate the training task multiple times through the training container under different driving scenarios to obtain multiple simulation results; a second evaluation unit, configured to evaluate the multiple simulation results in parallel and output the evaluation results.

[0094] Further references Figure 6 , which shows a structural schematic diagram of an embodiment of an autonomous driving training system according to the present disclosure.

[0095] like Figure 6 As shown, the autonomous driving training system 600 of this embodiment may include a client 601 and a cloud platform 602. The client 601 is configured to initiate an autonomous driving training request to the cloud platform 602; the cloud platform 602 is configured to receive the autonomous driving training request from the client 601, determine the training task specified in the autonomous driving training request, obtain a basic training image, and a training dataset and task code package corresponding to the training task; add the training dataset and task code package to the basic training image to obtain a training image; create a training container based on the training image, and practice the training task using the training container.

[0096] In this embodiment, the client 601 may first initiate a login request to the cloud platform 602. After receiving the login request, if the cloud platform 602 determines that the login request meets the login requirements, it allows the client 601 to log in to the cloud platform 602 and initiates an autonomous driving training request to the cloud platform. After receiving the autonomous driving training request from the client 601, the cloud platform 602 may perform the autonomous driving training method specifically as described in the following examples. Figure 2-Figure 4 The implementation of the method shown will not be described in detail here.

[0097] In some optional implementations of this embodiment, the autonomous driving training system 600 further includes: a proxy server; and a client 602, which is also used to obtain the remote terminal service, training container service, and online code editing service of the cloud platform 601 through the proxy server.

[0098] In this embodiment, the role of the proxy server is to act as an agent for the client to obtain information on the cloud platform. It can be regarded as a server between the client and the cloud platform. In this case, the client does not directly access the cloud platform but sends a request to the proxy server, and further obtains the remote terminal service, training container service and online code editing service provided by the cloud platform through the proxy server. Among them, the remote terminal service is used to authorize the client to remotely access the cloud platform online, the training container service is used to build and run the training container, and the online code editing service is used to receive the client's modification operations on the task code package. By setting up a proxy server, the access speed and security of the cloud platform can be improved.

[0099] Further references Figure 7 , which shows a schematic diagram of the resource architecture of the cloud platform in the autonomous driving training system disclosed herein. The cloud platform in this embodiment, its hardware devices may include GPU / CPU, storage resource pool, network resource pool and elastic resource expansion. Virtualized resource pools can be built on the hardware devices, including virtualized computing resource pools, storage resource pools and network resource pools. At the same time, Docker containers can be managed based on K8s (Kubernetes), a container management system for automatically deploying, scaling and managing containerized applications. Docker container technology is an open source application container engine. At the system function level on the virtualized resource pool, the autonomous driving training system of this application can be established, including an autonomous driving training scheduling system, a training evaluation system and a training monitoring system. The autonomous driving training scheduling system may include training modules such as positioning, high-precision mapping, perception, planning and control to implement various training functions; the training evaluation system may include breakpoint evaluation, result accuracy evaluation and driving safety evaluation, which can be used to comprehensively evaluate the training effect; the training monitoring system includes log monitoring, business monitoring and alarms, which can record specific information of each training. At the application level above the system function level, there are mainly two application scenarios, namely, employee training for autonomous driving companies and practical training in autonomous driving related professional courses in colleges and universities. These companies and professional personnel are the core force in the learning and development of autonomous driving functions, and are also the main audience of the autonomous driving cloud platform.

[0100] Figure 8This is a schematic diagram of the application process of the disclosed autonomous driving training method on a cloud platform. Students first need to log in to the autonomous driving training cloud platform through a client. The cloud platform authenticates the login request and allows login if the authentication is successful. Afterwards, some of the training services provided by the cloud platform are free, while others require payment or are targeted to specific individuals. Students need to redeem the relevant services through free redemption or points redemption. Next, the autonomous driving training scheduling system can obtain the corresponding autonomous driving dataset (i.e., training dataset) and autonomous driving code package (i.e., task code package) based on the received training request, as well as the autonomous driving training image (i.e., basic training image). By adding different datasets and code packages to the autonomous driving training image, different training modules can be obtained, including a positioning training module, a high-precision map training module, a perception training module, a decision-making and planning training module, a control training module, and a middleware communication training module. Subsequently, an autonomous driving training configuration can be generated based on the configuration information set by the student and the basic information of the image. A training container deployment is then generated based on the training configuration, and a training container is created based on the deployment. The client can access the online code editor through the proxy module to modify the autonomous driving code package. It can also access the cloud platform's remote terminal service and training container service through the proxy module to run and debug the training container. After the container is running, the evaluation and analysis module can perform a single evaluation of the results of a single run or conduct parallel evaluations of multiple runs. If the evaluation passes, an evaluation report is generated, and the training session ends. If the evaluation fails, the evaluation results are fed back to the autonomous driving training scheduling system, which reselects the required training tasks and repeats the training process.

[0101] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0102] Figure 9 A schematic block diagram of an example electronic device 900 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0103] like Figure 9As shown, the device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. Various programs and data required for the operation of the device 900 can also be stored in the RAM 903. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0104] Various components in the device 900 are connected to the I / O interface 905, including an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a magnetic disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0105] The computing unit 901 can be various general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above, such as the autonomous driving training method. For example, in some embodiments, the autonomous driving training method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 908. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the autonomous driving training method described above can be performed. Alternatively, in other embodiments, the computing unit 901 can be configured to perform the autonomous driving training method by any other appropriate means (e.g., by means of firmware).

[0106] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0107] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0108] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0109] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0110] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0111] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a server in a distributed system or a server integrated with blockchain. The server may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. The server may be a server in a distributed system or a server integrated with blockchain. The server may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0112] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0113] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0114] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for training an autonomous driving vehicle, comprising: receiving an autonomous driving training request and determining a training task specified in the autonomous driving training request; Obtaining a basic training image, as well as a training data set and a task code package corresponding to the training task, including: specifying multiple training tasks in response to the autonomous driving training request, and respectively obtaining a training data set and a task code package corresponding to each training task; Adding the training data set and the task code package into the basic training image to obtain the image to be trained, including: adding target training data into the basic training image to obtain the image to be trained, wherein the target training data is obtained by rearranging the obtained multiple training data sets and multiple task code packages; A training container is created based on the image to be trained, and the training task is practiced through the training container.

2. The method according to claim 1, wherein The training tasks include at least one of the following: positioning training tasks, high-precision map training tasks, perception training tasks, decision-making and planning training tasks, control training tasks and middleware communication training tasks.

3. The method according to claim 1, wherein The step of creating a training container based on the image to be trained includes: Get training configuration information; Generate container deployment information based on the training configuration information; A training container is created based on the container deployment information and the image to be trained.

4. The method according to claim 1, further comprising: receiving a modification operation on the task code package to obtain a user code package; The step of adding the training data set and the task code package into the basic training image to obtain the image to be trained includes: The training data set and the user code package are added to the basic training image to obtain the image to be trained.

5. The method according to claim 1, wherein practicing the training task through the training container comprises: According to the drill steps, the training task is drilled step by step through the training container; In response to the completion of execution of one of the drill steps, obtaining an output result corresponding to the drill step; The output result is evaluated and the evaluation result is output.

6. The method according to claim 1, wherein practicing the training task through the training container comprises: Under different driving scenarios, the training task is simulated multiple times by the training container to obtain multiple simulation results; The multiple simulation results are evaluated in parallel and the evaluation results are output.

7. An autonomous driving training device, comprising: a receiving module configured to receive an autonomous driving training request and determine a training task specified in the autonomous driving training request; An acquisition module is configured to acquire a basic training image, and a training data set and a task code package corresponding to the training task, including: responding to the autonomous driving training request to specify multiple training tasks, and respectively acquiring a training data set and a task code package corresponding to each training task; An adding module is configured to add the training data set and the task code package into the basic training image to obtain the image to be trained, including: adding target training data into the basic training image to obtain the image to be trained, wherein the target training data is obtained by rearranging the obtained multiple training data sets and multiple task code packages; The training module is configured to create a training container based on the image to be trained, and to practice the training task through the training container.

8. The device according to claim 7, wherein The training tasks include at least one of the following: positioning training tasks, high-precision map training tasks, perception training tasks, decision-making and planning training tasks, control training tasks and middleware communication training tasks.

9. The device according to claim 7, wherein The drill module includes: A second acquiring unit is configured to acquire training configuration information; A generating unit configured to generate container deployment information based on the training configuration information; The creation unit is configured to create a training container based on the container deployment information and the image to be trained.

10. The apparatus according to claim 7, further comprising: a modification module configured to receive a modification operation on the task code package and obtain a user code package; The adding module includes: The second adding unit is configured to add the training data set and the user code package into the basic training image to obtain the image to be trained.

11. The device according to claim 7, wherein The drill module includes: A rehearsal unit is configured to rehearse the training task step by step through the training container according to the rehearsal steps; A result acquisition unit is configured to acquire an output result corresponding to a drill step in response to the completion of the drill step; The first evaluation unit is configured to evaluate the output result and output the evaluation result.

12. The device according to claim 7, wherein The drill module includes: A simulation unit is configured to simulate the training task multiple times through the training container in different driving scenarios to obtain multiple simulation results; The second evaluation unit is configured to perform parallel evaluation on the multiple simulation results and output the evaluation results.

13. An autonomous driving training system, comprising: The client is used to initiate autonomous driving training requests to the cloud platform; A cloud platform, configured to receive the autonomous driving training request from the client and determine a training task specified in the autonomous driving training request; Obtaining a basic training image, as well as a training data set and a task code package corresponding to the training task, including: specifying multiple training tasks in response to the autonomous driving training request, and respectively obtaining a training data set and a task code package corresponding to each training task; adding the training data set and the task code package into the basic training image to obtain the image to be trained, including: adding target training data into the basic training image to obtain the image to be trained, wherein the target training data is obtained by re-arranging the obtained multiple training data sets and multiple task code packages; creating a training container based on the image to be trained, and rehearsing the training task through the training container.

14. The system of claim 13, further comprising: Proxy server; The client is also used to obtain the remote terminal service, training container service and online code editing service of the cloud platform through the proxy server.

15. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 6.

17. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images