Software architecture platform, control method, vehicle and medium of an autonomous vehicle

Abstract

This application relates to the technical field of autonomous driving application layer algorithms, and particularly relates to a software architecture platform, a control method, a vehicle, and a medium for an autonomous driving vehicle. Among them, the method includes: matching a preset driving algorithm or parking algorithm for the vehicle according to the running state information and decision-making scenarios of the vehicle, and predicting the target trajectory based on the local environment information and target information sensed by the vehicle; determining the reference trajectory, reference trajectory boundary, reference speed, and reference speed boundary for the vehicle's driving or parking according to the vehicle's body information, map, positioning information, and predicted trajectory; and generating the optimal trajectory and optimal speed for driving or parking based on the obtained reference values, controlling the vehicle to execute the preset driving or parking actions, and realizing the preset driving function and preset parking function of the vehicle. Thereby, the problems in the related art that the continuity of the daily travel scene experience cannot be achieved, the driving and parking software architectures cannot be shared, and the software scheduling efficiency is low are solved.

Description

Technical Field

[0001] This application relates to the technical field of autonomous driving application layer algorithms, and particularly relates to a software architecture platform, a control method, a vehicle and a medium for an autonomous driving vehicle. Background Art

[0002] Currently, the mainstream mass-produced autonomous driving vehicles generally have driving and parking functions. However, driving and parking are generally developed based on different architectures and deployed on different domain controllers. As a result, the driving and parking software architectures cannot be shared, the software modules cannot be shared, and the computing resources cannot be shared, resulting in a waste of hardware costs and an increase in development labor costs.

[0003] From the perspective of user experience, the current mainstream autonomous driving products mainly cover highway scenarios and garage parking scenarios, and cannot achieve a continuous experience from the home garage to the company garage or from any point A to point B, and do not support software self-evolution. Summary of the Invention

[0004] This application provides a software architecture platform, a control method, a vehicle and a medium for an autonomous driving vehicle to solve the problems in the related technologies, such as the inability to achieve the continuity of daily travel scenario experience, the inability to share the driving and parking software architectures, and the low software scheduling efficiency.

[0005] The first aspect of the embodiments of the present application provides a software architecture platform for an autonomous vehicle. The vehicle is provided with preset driving functions and preset parking functions. The platform is used to control the vehicle to implement the preset driving functions and the preset parking functions, including: an acquisition module, configured to acquire the local environment information perceived by the vehicle, the time-series information of the target or obstacle, as well as the operating state information and decision-making scenarios of the vehicle; a prediction module, configured to call a prediction algorithm that meets the current scenario according to the operating state information and decision-making scenarios, and perform trajectory prediction on the local environment information and the time-series information of the target or obstacle based on the prediction algorithm to generate a predicted trajectory of the targets around the vehicle; a decision-making module, configured to call a decision-making algorithm that meets the current scenario according to the decision-making scenarios, make a decision based on the decision-making algorithm on the vehicle body information, map, positioning information and the predicted trajectory to obtain a reference trajectory, reference trajectory boundary, reference speed and reference speed boundary for driving or parking; a planning module, configured to call a planning algorithm that meets the current scenario according to the decision-making scenarios, use the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate an optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate an optimal speed for driving or parking; a control module, configured to call a control algorithm that meets the current scenario according to the decision-making scenarios, and use the control algorithm to control the vehicle to execute preset driving actions or preset parking actions according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

[0006] According to the above technical means, the embodiments of the present application can call a prediction algorithm suitable for the current scenario according to the vehicle system operating state and decision-making scenarios, and combine the vehicle environment information and the time-series information of the target or obstacle to generate a predicted trajectory of the targets around the vehicle, and use information such as vehicle body information and map positioning to plan the optimal trajectory and speed for the vehicle to drive or park, so as to realize the automatic driving and / or automatic parking of the vehicle. By adopting an integrated driving and parking regulation and control architecture, the same set of software architecture is used for driving and parking. By calling the corresponding algorithm according to the scenario, the architecture can be shared and the modules can be shared, reducing the consumption of computing resources, reducing the number and cost of controllers, reducing the labor cost, and can realize that the platform algorithm is applicable to controllers and vehicle models with different computing powers through configuration files, realizing a platformized development mode and improving the development efficiency.

[0007] Optionally, in an embodiment of the present application, a scene recognition module is provided to determine the current decision-making scene according to the vehicle driving environment. The decision-making scene includes one or more of a lane following scene, an intersection scene, a garage driving scene, a garage parking scene, and a U-turn scene. These scenes belong to driving decision-making scenarios and parking decision-making scenarios. The driving decision-making scenarios include a lane following scene, an intersection scene, a garage driving scene, and a U-turn scene. The parking decision-making scenario includes a garage parking scene. Among them,

[0008] If the vehicle meets the preset U-turn scene condition, the vehicle enters the U-turn scene, and then calls the prediction algorithm, decision algorithm, planning algorithm, and control algorithm required for the U-turn scene;

[0009] If the vehicle meets the preset intersection scene condition, the vehicle enters the intersection scene, and then calls the prediction algorithm, decision algorithm, planning algorithm, and control algorithm required for the intersection scene;

[0010] If the vehicle meets the preset garage driving scene condition, the vehicle enters the garage scene, and then calls the prediction algorithm, decision algorithm, planning algorithm, and control algorithm required for the garage driving scene;

[0011] If the vehicle meets the preset parking scene condition, the vehicle enters the garage parking scene, and then calls the prediction algorithm, decision algorithm, planning algorithm, and control algorithm required for the U-turn parking scene.

[0012] According to the above technical means, the decision-making scenes in the embodiment of the present application are a lane following scene, an intersection scene, a garage driving scene, a garage parking scene, and a U-turn scene. When a certain condition is met, the vehicle enters the corresponding scene, and the optimal algorithm is called through scene refinement, improving the continuity of the daily travel scene experience.

[0013] Optionally, in an embodiment of the present application, the decision module includes: a processing sub-module, which is used to determine a candidate lane set based on the actual type of the map, screen the candidate lanes in the candidate lane set that meet the preset conditions, determine a candidate coordinate system and boundary according to the candidate lanes and a preset reference line, make a traffic rule decision according to the candidate coordinate system and boundary to obtain a traffic rule decision result, and mark the obstacles around the vehicle to obtain the horizontal and vertical marking results of the obstacles; a decision sub-module, which is used to call the decision algorithm that meets the current scene according to the decision-making scene, use the decision algorithm to make a decision on the traffic rule decision result and the horizontal and vertical marking results of the obstacles, determine the path boundary constraint and speed boundary constraint in the candidate coordinate system, and perform trajectory planning and trajectory feature extraction according to the path boundary constraint and speed boundary constraint to obtain the reference trajectory, reference trajectory boundary, reference speed, and reference speed boundary for the driving or parking of the vehicle.

[0014] According to the above technical means, the embodiment of the present application can output the reference trajectory and reference speed of the vehicle for driving or parking according to the decision-making module of the vehicle, and obtain the left and right boundary values respectively, and realize the path planning of the vehicle for driving or parking by narrowing the reference range.

[0015] Optionally, in an embodiment of the present application, the processing sub-module is specifically configured to: if the map is a preset high-precision map, process the preprocessed data by using a strategy of correcting the preset high-precision map and the perception information; if the map is a preset crowdsourced map, process the preprocessed data by using a strategy of combining the preset crowdsourced map and the perception information, where the preset crowdsourced map is a road prior information map generated by combining traffic elements sensed in multiple times.

[0016] According to the above technical means, the environmental model sub-module of the embodiment of the present application can process the candidate lane output when the map sources are different, and based on the perception crowdsourcing mapping, realize the connection of the open road and the garage road, and realize the continuous experience of driving and parking integration.

[0017] Optionally, in an embodiment of the present application, the strategy of combining the preset crowdsourced map and the perception information includes: judging the lane information of the lane where the vehicle is currently located according to the preprocessed data; performing positioning verification according to the lane information and the perception information to determine the lane identifier of the lane where the vehicle is currently located, searching for the navigation reference center line by using the lane identifier to obtain a reference path sequence, and determining the position relationship of the reference path center line points based on the reference path sequence; performing mutual verification and fusion according to the perception information and the position relationship of the reference path center line points to obtain the left and right lane line relationships of the currently located lane, performing mutual verification and fusion according to the perception information and the left and right lane line relationships to obtain lane attribute expansion information; performing mutual verification and fusion according to the perception information, the lane attribute expansion information, the reference line center line and the boundary line point sequence, outputting the reference center line and the boundary sequence, and determining the candidate lane set based on the reference center line and the boundary sequence.

[0018] According to the above technical means, the embodiment of the present application can quickly update the difference between the map and the real environment by building a map through perception information during the actual road update, and ensure the continuous performance when the crowdsourced map fails temporarily or the elements are abnormal through multiple verification and fusion.

[0019] Optionally, in an embodiment of the present application, the prediction module includes: a data preprocessing sub-module, configured to preprocess the environmental information and the target or obstacle time-series information to obtain target information and preprocessed environmental information; a task scheduling sub-module, configured to call a prediction algorithm that meets the current scenario according to the operation state information and the decision-making scenario; a prediction algorithm sub-module, configured to perform destination prediction, intention prediction, and deep learning-based trajectory prediction on the target information and the preprocessed environmental information to obtain a predicted destination, a target intention, and a model-generated trajectory, and obtain a rule-generated trajectory based on the predicted destination and the target intention, and perform trajectory processing on the model-generated trajectory and the rule-generated trajectory to obtain a predicted trajectory of the target around the vehicle.

[0020] According to the above technical means, the embodiment of the present application can obtain a predicted trajectory of the target around the vehicle according to the local environmental information, target or obstacle time-series information sensed by the vehicle, as well as the operation state information and decision-making scenario of the vehicle, and can further determine the driving or parking operation trajectory and speed of the vehicle in combination with the vehicle body information, map, and positioning information.

[0021] Optionally, in an embodiment of the present application, the task scheduling sub-module is specifically configured to call a preset driving prediction algorithm if the vehicle is in a preset lane-following scenario, where when the distance between the vehicle and the garage exit in the driving direction is less than a first threshold, the vehicle enters the preset lane-following scenario; if the vehicle is in a preset garage driving scenario, call a preset garage driving scenario prediction algorithm, where when the distance between the vehicle and the garage entrance in the driving direction is less than a second threshold, the vehicle enters the preset garage driving scenario.

[0022] According to the above technical means, the embodiment of the present application can determine the driving scenario of the vehicle according to the driving direction of the vehicle and the distance from the garage entrance, so as to call different driving prediction algorithms to achieve a continuous user experience.

[0023] Optionally, in an embodiment of the present application, the prediction module further includes: a data trigger sub-module, configured to trigger the recording of preset scenario data and problem data, and upload the triggered data to the server to achieve data closed-loop.

[0024] According to the above technical means, the embodiment of the present application can verify the real vehicle algorithm and analyze the data in a data closed-loop manner, improving the accuracy and applicability of the algorithm.

[0025] Optionally, in an embodiment of the present application, the planning module includes: a trajectory planning sub-module for planning an optimal path according to the reference trajectory, the reference trajectory boundary, the predicted trajectory, and the positioning information of the vehicle; a speed planning sub-module for planning an optimal speed according to the optimal path, the reference speed, and the reference speed boundary.

[0026] According to the above technical means, the embodiments of the present application can plan the motion trajectory and driving speed of the vehicle during driving or parking to achieve a continuous experience from the home garage to the company garage or from any point A to point B.

[0027] Optionally, in an embodiment of the present application, the control module adopts a control architecture combining feedforward and PID feedback, and the control architecture adopts a dual-loop PID control algorithm based on dynamically changing parameters.

[0028] According to the above technical means, the embodiments of the present application can implement functions in different scenarios through an integrated driving and parking control architecture and algorithm, improving the development efficiency.

[0029] An embodiment of the second aspect of the present application provides a control method for an autonomous vehicle. The method is applied to the vehicle to control the vehicle to implement preset driving functions and preset parking functions, and includes the following steps: obtaining the local environment information, target or obstacle time-series information sensed by the vehicle, as well as the operating state information and decision-making scenario of the vehicle; calling a prediction algorithm that meets the current scenario according to the operating state information and decision-making scenario, and based on the prediction algorithm, performing trajectory prediction on the local environment information and the target or obstacle time-series information to generate a predicted trajectory of the targets around the vehicle; calling a decision-making algorithm that meets the current scenario according to the decision-making scenario, and making a decision on the vehicle body information, map, positioning information, and the predicted trajectory based on the decision-making and planning algorithm to obtain a reference trajectory, reference trajectory boundary, reference speed, and reference speed boundary for driving or parking; calling a planning algorithm that meets the current scenario according to the decision-making scenario, using the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate an optimal trajectory for driving or parking, and using the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate an optimal speed for driving or parking; calling a control algorithm that meets the current scenario according to the decision-making scenario, and using the control algorithm to control the vehicle to execute preset driving actions or preset parking actions according to the optimal trajectory and the optimal speed, so as to achieve automatic driving and / or automatic parking of the vehicle.

[0030] A third aspect embodiment of the present application provides a vehicle, including an autonomous driving system; a software architecture platform for autonomous driving vehicles. Among them, the preset driving functions and preset parking functions of the autonomous driving system share the platform. The platform is used to obtain the environmental information, target information, operating state information, and decision-making scenarios around the vehicle; call a prediction algorithm that meets the current scenario according to the operating state information and decision-making scenario, and based on the prediction algorithm, perform trajectory prediction on the local environmental information and the target or obstacle time-series information to generate a predicted trajectory of the target around the vehicle; call a decision-making algorithm that meets the current scenario according to the decision-making scenario, and based on the decision-making and planning algorithm, make decisions on the vehicle body information, map, positioning information, and the predicted trajectory of the vehicle to obtain a reference trajectory, reference trajectory boundary, reference speed, and reference speed boundary for driving or parking; call a planning algorithm that meets the current scenario according to the decision-making scenario, use the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate an optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate an optimal speed for driving or parking; call a control algorithm that meets the current scenario according to the decision-making scenario, and use the control algorithm to control the vehicle to execute preset driving actions or preset parking actions according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

[0031] A fourth aspect embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and the program is executed by a processor to be used to implement the control method of the autonomous driving vehicle as described in the above embodiments.

[0032] Therefore, the present application has at least the following beneficial effects:

[0033] 1. Embodiments of the present application can call a prediction algorithm suitable for the current scenario according to the operating state of the vehicle system and the decision-making scenario, and combine the environmental information of the vehicle and the time-series information of the target or obstacle to predict the trajectory of the target around the vehicle. Use the vehicle body information, map positioning and other information to plan the optimal trajectory and speed for the vehicle to drive or park, so as to realize the automatic driving and / or automatic parking of the vehicle. By adopting an integrated driving and parking regulation and control architecture, the same set of software architecture is used for driving and parking. By calling the corresponding algorithm according to the scenario, the architecture sharing and module sharing can be realized, reducing the consumption of computing resources, reducing the number of controllers and costs, reducing the labor cost, and can realize that the platform algorithm is applicable to controllers and vehicle models with different computing powers through configuration files, realizing the platform development mode and improving the development efficiency.

[0034] 2. The decision-making scenarios of the embodiments of the present application include lane following scenarios, intersection scenarios, garage driving scenarios, garage parking scenarios, and U-turn scenarios. When a certain condition is met, the vehicle enters the corresponding scenario, and the optimal algorithm is called through scenario refinement, improving the continuity of the daily travel scenario experience.

[0035] 3. The embodiments of the present application can output the reference trajectory and reference speed of the vehicle's driving or parking according to the decision-making module of the vehicle, and obtain the left and right boundary values respectively. By narrowing the reference range, the path planning of the vehicle's driving or parking is realized.

[0036] 4. The environmental model sub-module of the embodiments of the present application can process the candidate lane output when different map sources are used, and based on perception crowdsourcing mapping, realize the connection of open roads and garage roads, and achieve the continuous experience of driving and parking integration.

[0037] 5. The embodiments of the present application can quickly update the difference between the map and the real environment by building a map through perception information during actual road updates. Through multiple verification fusions, it can ensure continuous performance when the crowdsourcing map fails temporarily or elements are abnormal.

[0038] 6. The embodiments of the present application can obtain the predicted trajectory of the targets around the vehicle according to the local environmental information, target or obstacle time-series information perceived by the vehicle, as well as the operating state information and decision-making scenarios of the vehicle. Combining the vehicle's body information, map, and positioning information, the driving or parking operation trajectory and speed of the vehicle can be further determined.

[0039] 7. The embodiments of the present application can judge the driving scenario of the vehicle according to the driving direction of the vehicle and the distance between the vehicle and the garage entrance, and thus call different driving prediction algorithms to achieve the continuous experience of users.

[0040] 8. The embodiments of the present application can verify the real vehicle algorithm and analyze the data in a data closed-loop manner, improving the accuracy and applicability of the algorithm.

[0041] 9. The embodiments of the present application can plan the motion trajectory and driving speed of the vehicle's driving or parking to achieve the continuous experience from the home garage to the company garage or from any point A to point B.

[0042] 10. The embodiments of the present application can realize the functions in different scenarios through the driving and parking integrated control architecture and algorithms, improving the development efficiency.

[0043] The additional aspects and advantages of the present application will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of embodiments in conjunction with the accompanying drawings, where:

[0045] Figure 1 FIG. 4 is a block diagram of a software architecture platform for an autonomous vehicle according to an embodiment of the present application;

[0046] Figure 2 FIG. 8 is an algorithm architecture diagram of a prediction module according to an embodiment of the present application;

[0047] Figure 3 FIG. 12 is a schematic diagram of the principle of switching between driving and parking according to an embodiment of the present application;

[0048] Figure 4 FIG. 16 is a data closed-loop architecture diagram according to an embodiment of the present application;

[0049] Figure 5 FIG. 20 is a decision scenario architecture diagram according to an embodiment of the present application;

[0050] Figure 6 FIG. 24 is an algorithm architecture diagram of a decision-making module according to an embodiment of the present application;

[0051] Figure 7 FIG. 28 is a reference path generation architecture diagram based on crowd-sourced maps according to an embodiment of the present application;

[0052] Figure 8 FIG. 32 is an algorithm architecture diagram of a planning module according to an embodiment of the present application;

[0053] Figure 9 FIG. 36 is an algorithm architecture diagram of a control module according to an embodiment of the present application;

[0054] Figure 10 FIG. 40 is a schematic flowchart of a control method for an autonomous vehicle according to an embodiment of the present application;

[0055] Figure 11 FIG. 44 is a block diagram of a vehicle according to an embodiment of the present application.

[0056] Explanation of reference numerals: Acquisition module - 100, Prediction module - 200, Decision-making module - 300, Planning module - 400, Control module - 500, Autonomous driving system - 600, Software architecture platform of autonomous vehicle - 10. Detailed Description of Embodiments

[0057] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present application and should not be construed as limiting the present application.

[0058] The software architecture platform, control method, vehicle, and medium of an autonomous vehicle according to an embodiment of the present application will be described below with reference to the accompanying drawings. In view of the problems mentioned in the above background art, the present application provides a software architecture platform for an autonomous vehicle. In this platform, according to the operating state of the vehicle system and the decision-making scenario, a prediction algorithm suitable for the current scenario is called, and the predicted trajectory of the target around the vehicle is generated by combining the environmental information of the vehicle and the temporal information of the target or obstacle. The optimal trajectory and speed of the vehicle for driving or parking are planned by using the body information of the vehicle, map positioning, and other information, so as to achieve automatic driving and / or automatic parking of the vehicle. By adopting an integrated driving and parking control architecture, the same set of software architecture is used for driving and parking. By calling the corresponding algorithm according to the scenario, the architecture can be shared and the modules can be shared, reducing the consumption of computing resources, reducing the number of controllers and costs, reducing labor costs, and realizing that the platform algorithm can be applied to controllers and vehicle models with different computing powers through configuration files, realizing a platform-based development mode and improving development efficiency. Thus, the problems in the related art that the continuity of the daily travel scenario experience cannot be achieved, software self-evolution cannot be supported, and the software scheduling efficiency is low are solved.

[0059] Specifically, Figure 1 is a block diagram of a software architecture platform for an autonomous vehicle provided by an embodiment of the present application.

[0060] As Figure 1 shown, the software architecture platform 10 of the autonomous vehicle includes: an acquisition module 100, a prediction module 200, a decision module 300, a planning module 400, and a control module 500.

[0061] Among them, the acquisition module 100 is used to acquire the local environment information perceived by the vehicle, the temporal information of the target or obstacle, as well as the operating state information and decision-making scenarios of the vehicle; the prediction module 200 is used to call the prediction algorithm that meets the current scenario according to the operating state information and decision-making scenarios, and perform trajectory prediction on the local environment information and the temporal information of the target or obstacle based on the prediction algorithm to generate the predicted trajectory of the target around the vehicle; the decision-making module 300 is used to call the decision-making algorithm that meets the current scenario according to the decision-making scenarios, and make decisions on the vehicle body information, map, positioning information and predicted trajectory based on the decision-making algorithm to obtain the reference trajectory, reference trajectory boundary, reference speed and reference speed boundary for driving or parking; the planning module 400 is used to call the planning algorithm that meets the current scenario according to the decision-making scenarios, use the planning algorithm to perform trajectory planning on the reference trajectory and reference trajectory boundary to generate the optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and reference speed boundary to generate the optimal speed for driving or parking; the control module 500 is used to call the control algorithm that meets the current scenario according to the decision-making scenarios, and use the control algorithm to control the vehicle to execute the preset driving action or preset parking action according to the optimal trajectory and optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

[0062] It can be understood that the vehicle in the embodiment of the present application is provided with a preset driving function and a preset parking function. The software architecture platform of the autonomous vehicle is used to control the vehicle to realize the preset driving function and the preset parking function. Through modules such as prediction, decision-making, planning, and control, task scheduling based on scenarios can be realized, and flexible scheduling of software modules can be achieved; and by calling the corresponding algorithms for different scenarios, functions in different scenarios can be realized, improving the continuity of the daily travel scenario experience.

[0063] When designing the architecture, the embodiment of the present application can set the operation cycle of each module according to the actual situation without specific limitation. For example, the operation cycle of the prediction module 200 can be set to 40 ms, the operation cycle of the decision-making module 300 can be set to 50 ms, the operation cycle of the planning module 400 can be set to 50 ms, and the operation cycle of the control module 500 can be set to 20 ms. At the same time, the allocation of computing resources in the embodiment of the present application can also be set according to the actual situation. As an example, it is specifically as follows:

[0064] 1) The resource occupancy rate of the planning and control CPU (central processing unit) is set to 30% of the total computing power, and the memory occupancy rate is set to 20% of the total memory; the AI computing power of the GPU (graphics processing unit) is set to 15%;

[0065] 2) The CPU occupancy rate of the prediction module is 10%, the memory occupancy rate is 8%, and the GPU AI computing power is set to 15%.

[0066] 3) The CPU occupancy rate of the decision-making module is 10%, the memory occupancy rate is 5%, and the GPU AI computing power is set to 0.

[0067] 4) The CPU occupancy rate of the planning module is 8%, the memory occupancy rate is 5%, and the GPU AI computing power is set to 0.

[0068] 5) The CPU occupancy rate of the control module is 2%, the memory occupancy rate is 2%, and the GPU AI computing power is set to 0.

[0069] In the embodiment of the present application, the decision-making scenarios include one or more of a lane-following scenario, an intersection scenario, a garage driving scenario, a garage parking scenario, and a U-turn scenario. These scenarios belong to driving decision-making scenarios and parking decision-making scenarios. The driving decision-making scenarios include a lane-following scenario, an intersection scenario, a garage driving scenario, and a U-turn scenario. The parking decision-making scenarios include a garage parking scenario, and its architecture is as Figure 2 shown.

[0070] Among them, the U-turn scenario includes: a scenario of approaching the U-turn area and a scenario of entering the U-turn state;

[0071] The intersection scenario includes: approaching the intersection and cruising at the intersection;

[0072] The garage parking scenario includes: approaching the parking space and starting to park;

[0073] The garage driving scenario includes: entering the garage scenario and exiting the garage scenario. Among them, the scenario of entering the garage includes: approaching the garage entrance, passing through the garage entrance, and lane following; the scenario of exiting the garage includes: approaching the garage exit, passing through the garage entrance, and lane following.

[0074] In the embodiment of the present application, by setting up a scenario recognition module, according to the vehicle driving environment, the current decision-making scenario is judged. The system default scenario can be set to the lane-following scenario. When the vehicle meets the preset U-turn scenario conditions, the vehicle is controlled to enter the U-turn scenario; when the vehicle meets the preset intersection scenario conditions, the vehicle is controlled to enter the intersection scenario; when the vehicle meets the preset garage driving scenario conditions, the vehicle is controlled to enter the garage scenario; when the vehicle meets the preset parking scenario conditions, the vehicle is controlled to enter the garage parking scenario. When entering the corresponding scenario, the optimal algorithm required by the scenario is called through scenario refinement, improving the continuity of the daily travel scenario experience.

[0075] In an embodiment of the present application, the prediction module 200 includes: a data preprocessing sub-module, a task scheduling sub-module, and a prediction algorithm sub-module.

[0076] Among them, the data preprocessing sub-module is used to preprocess the environmental information and the target or obstacle time-series information to obtain the target information and the preprocessed environmental information; the task scheduling sub-module is used to call the preset driving algorithm or the preset parking algorithm that meets the current scenario according to the running state information and the decision-making scenario; the prediction algorithm sub-module is used to perform destination prediction, intention prediction, and deep learning-based trajectory prediction based on the preset driving algorithm or the preset parking algorithm, the target information, and the preprocessed environmental information to obtain the predicted destination, the target intention, and the model-generated trajectory, and obtain the rule-generated trajectory based on the predicted destination and the target intention, and perform trajectory processing on the model-generated trajectory and the rule-generated trajectory to obtain the predicted trajectory of the target around the vehicle.

[0077] The input of the prediction module 200 in the embodiment of the present application includes: the post-fusion target, the front view target, the front view undistorted image, the positioning, the FreeSpace (driveable area), and the static map information. The output of the prediction module includes: the data trigger signal, the traffic participant intention, and the traffic participant trajectory. The prediction module 200 in the embodiment of the present application can obtain the predicted intention result and the predicted trajectory result of the vehicle through these three sub-modules: the data preprocessing sub-module, the task scheduling sub-module, and the prediction algorithm sub-module.

[0078] Specifically, as Figure 3 shown, the data preprocessing sub-module in the embodiment of the present application includes the preprocessing of the target or obstacle time-series information and the preprocessing of the environmental model information. Among them, the target or obstacle time-series information may include: the target (post-fusion) target (front view), the front view (undistorted), and the positioning information, etc.; the environmental information may include: the driveable area grid, the static map information, and the positioning information, etc. The embodiment of the present application preprocesses the local environmental information, the target or obstacle time-series information sensed by the vehicle through the data preprocessing sub-module. The task scheduling sub-module in the embodiment of the present application mainly calls the prediction algorithm suitable for the current scenario according to the system running state and the downstream decision-making scenario; the prediction algorithm sub-module in the embodiment of the present application can use the prediction algorithm of the current scenario and the processed environmental information and the target or obstacle time-series information to perform destination prediction and intention prediction to obtain the A1 intention prediction (planning or machine learning), the A2 destination prediction, and further obtain the rule-generated trajectory; it can also obtain the model-generated trajectory according to the A4 deep learning trajectory prediction, and through the A5 processing of the rule-generated trajectory and the model-generated trajectory, thereby obtaining the trajectory prediction result of the vehicle. At the same time, the intention prediction result can also be obtained according to the A1 intention prediction (planning or machine learning).

[0079] In one embodiment of the present application, the task scheduling sub-module is specifically configured to, if the vehicle is in a preset lane following scenario, call a preset driving prediction algorithm, where when the distance between the vehicle and the garage exit in the driving direction is less than a first threshold, it enters the preset lane following scenario; if the vehicle is in a preset garage driving scenario, call a preset garage driving scenario prediction algorithm, where when the distance between the vehicle and the garage entrance in the driving direction is less than a second threshold, it enters the preset garage driving scenario.

[0080] As Figure 4 shown, the integrated driving and parking prediction algorithm of the embodiment of the present application is implemented through the task scheduling module; when the vehicle is in the lane following scenario, the driving prediction algorithm is called through the task scheduling module; when the vehicle is in the garage driving scenario, the garage driving scenario prediction algorithm is called through the task scheduling module. When the distance between the vehicle and the garage entrance in the driving direction is less than a first threshold, it enters the garage driving scenario; when the distance between the vehicle and the garage exit in the driving direction is less than a second threshold, it enters the lane following scenario. Among them, the first threshold and the second threshold can be determined according to the actual situation and are not specifically limited.

[0081] In one embodiment of the present application, the prediction module 200 further includes: a data triggering sub-module. Among them, the data triggering sub-module is used to trigger the recording of preset scenario data and problem data, and upload the triggered data to the server to achieve data closed-loop.

[0082] The embodiment of the present application can output a data trigger signal through the data trigger module. As Figure 5 shown, the data closed-loop of the embodiment of the present application mainly includes the following steps: training data collection; data cleaning to generate structured data; general feature extraction; sample extraction; algorithm training; in-vehicle deployment and optimization; in-vehicle algorithm verification; data recording; problem data analysis.

[0083] After the data analysis is completed, if it is a data-driven problem, the data is merged into the original data; if it is not a data-driven problem, the following steps are entered: data reinjection analysis; algorithm update; data reinjection verification; in-vehicle deployment and optimization.

[0084] In one embodiment of the present application, the decision module 300 includes: a processing sub-module and a decision sub-module,

[0085] Among them, the processing sub-module is used to determine a candidate lane set based on the actual type of the map, screen candidate lanes that meet preset conditions in the candidate lane set, determine a candidate coordinate system and boundaries based on the candidate lanes and a preset reference line, make a preset traffic rule decision according to the candidate coordinate system and boundaries to obtain a traffic rule decision result, and mark obstacles around the vehicle to obtain horizontal and vertical marking results of the obstacles; the decision sub-module is used to call a decision algorithm that meets the current scenario according to the decision scenario, use the decision algorithm to make a decision on the traffic rule decision result and the horizontal and vertical marking results of the obstacles, determine path boundary constraints and speed boundary constraints in the candidate coordinate system, and perform trajectory planning and trajectory feature extraction according to the path boundary constraints and speed boundary constraints to obtain a reference trajectory, a reference trajectory boundary, a reference speed, and a reference speed boundary for the vehicle to drive or park.

[0086] Among them, the decision-making environment model in the embodiments of the present application includes: a model with a high-precision map as the main and perception information as the auxiliary, and a model that combines a crowdsourced map and perception information.

[0087] Specifically, the decision-making module 300 in the embodiments of the present application can be detailedly divided into: a data preprocessing sub-module, an environment model sub-module, a candidate lane sub-module, a reference line smoothing sub-module, a traffic rule decision sub-module, an obstacle decision sub-module, a decision scenario sub-module, a task planning sub-module, and a post-decision sub-module. As Figure 6 shown, in the embodiments of the present application, by inputting vehicle body information, a map, positioning information, and a predicted trajectory, and performing operations such as screening, processing, evaluation, and extraction through multiple sub-modules in the decision-making module 200, the optimal trajectory for the vehicle to drive or park can be output. Among them, the optimal trajectory includes: a reference trajectory and its left and right boundaries, a reference speed, and upper and lower speed boundaries. By narrowing the reference range, path planning for the vehicle to drive or park is realized.

[0088] In an embodiment of the present application, the processing sub-module is specifically used to, if the map is a preset high-precision map, process the preprocessed data using a strategy of correcting the preset high-precision map and perception information; if the map is a preset crowdsourced map, process the preprocessed data using a strategy of combining the preset crowdsourced map and perception information, where the preset crowdsourced map is a road prior information map generated by combining traffic elements sensed in multiple real-time manners.

[0089] It can be understood that the environment model sub-module in the embodiments of the present application is used to process the output of candidate lanes when the map source is different. When the map source is a high-precision map, the data is processed using a strategy with the high-precision map as the main and perception information for correction; when the map source is a crowdsourced map, the data is processed using a strategy of combining the crowdsourced map and perception information. Through perception-based crowdsourced mapping, the connection between open roads and garage roads is realized, and the continuous experience of driving and parking in one is achieved;

[0090] In the actual execution process, the crowdsourced map combines traffic elements sensed in real time multiple times. In the embodiments of the present application, a road prior information map can be generated through an automated mapping tool.

[0091] In an embodiment of the present application, a preset strategy for combining the crowdsourced map and the sensed information includes: judging the lane information of the lane where the vehicle is currently located according to the preprocessed data; performing positioning verification according to the lane information and the sensed information to determine the lane identifier of the lane where the vehicle is currently located, searching for the navigation reference center line using the lane identifier to obtain a reference path sequence, and determining the positional relationship of the reference path center line points based on the reference path sequence; performing mutual verification and fusion according to the sensed information and the positional relationship of the reference path center line points to obtain the left and right lane line relationships of the lane where the vehicle is currently located, and performing mutual verification and fusion according to the sensed information and the left and right lane line relationships to obtain the lane attribute expansion information; performing mutual verification and fusion according to the sensed information, the lane attribute expansion information, the reference line center line and the boundary line point sequence, outputting the reference center line and the boundary sequence, and determining the candidate lane set based on the reference center line and the boundary sequence.

[0092] Specifically, as Figure 7 shown, the strategy for combining the crowdsourced map and the sensed information in the embodiments of the present application includes: judging the currently occupied lane, positioning verification, searching for the navigation reference center line, confirming the positional relationship of the reference path center line points, judging the left and right adjacent lanes, element expansion and front and back continuation, reconstructing and matching the center line and the boundary line, and finally outputting the reference center line and the reference boundary sequence. In the embodiments of the present application, the candidate lane set can be determined through the reference center line and the boundary sequence. When actually updating the road, in the embodiments of the present application, the map can be built through the sensed information to quickly update the difference between the map and the real environment. Through multiple verification and fusion, the continuity performance can be maintained when the crowdsourced map fails short-term or the elements are abnormal.

[0093] Among them, the input of the positioning verification module includes the information of the currently occupied lane and the sensed information A, and the sensed information A includes traffic elements, obstacles, stop lines, lane lines, guardrails, vehicle flows, and freespace;

[0094] The input of the left and right adjacent lane judgment module includes the sensed information A and the positional relationship of the reference path center line points, and the output is the left and right lane line relationship, and mutual verification and fusion are performed between the two;

[0095] The input of the element left and right expansion and front and back continuation includes the sensed information A, the reference path center line association relationship, and the left and right lane relationships, and mutual verification and fusion are performed among the three; the output is the lane attribute expansion;

[0096] The input of the centerline and boundary line reconstruction and matching module includes lane attribute expansion, perception information A, and the point sequences of the reference line centerline and boundary line. Mutual verification and fusion are performed among the three, and the output is the reference centerline and boundary sequence.

[0097] In one embodiment of the present application, the planning module 400 includes: a trajectory planning sub-module for planning an optimal path according to the reference trajectory, reference trajectory boundary, predicted trajectory, and vehicle positioning information; and a speed planning sub-module for planning an optimal speed according to the optimal path, reference speed, and reference speed boundary.

[0098] As Figure 8 shown, the planning module 400 of the embodiment of the present application can input the decision reference centerline (including the reference trajectory and its left and right boundaries, the reference speed and its upper and lower boundaries), the target predicted trajectory, and the positioning information. Through the pre-processing, rough search, and digital optimization processing methods of the trajectory planning sub-module and the speed planning sub-module, the optimal trajectory and optimal speed for vehicle driving or parking can be obtained. At the same time, the embodiment of the present application transmits the planned path to the speed planning sub-module and transmits the planned speed result to the trajectory planning sub-module, so as to achieve the intercommunication and resource sharing between the sub-modules, in order to output the best control trajectory required by the vehicle, including the centerline and speed.

[0099] In one embodiment of the present application, the control module 500 adopts a control architecture combining feedforward and PID feedback, and the control architecture adopts a dual-loop PID control algorithm based on dynamically changing parameters.

[0100] It can be understood that the embodiment of the present application can share the algorithm library including static and dynamic architectures and a common standardized interface for driving and parking, and call the optimal algorithm through scenario refinement. As Figure 9 shown, it adopts an improved feedforward plus PID feedback control architecture and a dual-loop PID control algorithm based on dynamically changing parameters.

[0101] The software architecture platform of an autonomous vehicle proposed according to an embodiment of the present application calls a prediction algorithm suitable for the current scenario according to the operating state of the vehicle system and the decision-making scenario, and generates a predicted trajectory of the target around the vehicle by combining the environmental information of the vehicle and the time-series information of the target or obstacle. The optimal trajectory and speed of the vehicle for driving or parking are planned by using the vehicle body information, map positioning and other information of the vehicle, so as to realize the automatic driving and / or automatic parking of the vehicle. By adopting an integrated driving and parking control architecture, the same set of software architecture is used for driving and parking. By calling the corresponding algorithm according to the scenario, the architecture can be shared and the modules can be shared, reducing the consumption of computing resources, reducing the number of controllers and costs, reducing labor costs, and realizing that the platform algorithm can be applied to controllers and vehicle models with different computing powers through configuration files, realizing a platformized development mode and improving development efficiency. Thus, the problems that the related technology cannot realize the continuity of the daily travel scenario experience, and the driving and parking software architectures cannot be shared and the software scheduling efficiency is low are solved.

[0102] Next, a control method for an autonomous vehicle proposed according to an embodiment of the present application is described with reference to the accompanying drawings. The method is applied to a vehicle and is used to control the vehicle to realize a preset driving function and a preset parking function.

[0103] Figure 10 It is a flowchart of a control method for an autonomous vehicle provided by an embodiment of the present application.

[0104] As Figure 10 shown, the control method of the autonomous vehicle includes the following steps:

[0105] In step S101, the local environmental information, the time-series information of the target or obstacle, and the operating state information and decision-making scenario of the vehicle sensed by the vehicle are obtained;

[0106] In step S102, a prediction algorithm that meets the current scenario is called according to the operating state information and the decision-making scenario, and the local environmental information and the time-series information of the target or obstacle are subjected to trajectory prediction based on the prediction algorithm to generate a predicted trajectory of the target around the vehicle;

[0107] In step S103, a decision algorithm that meets the current scenario is called according to the decision-making scenario, and decisions are made on the vehicle body information, map, positioning information and predicted trajectory of the vehicle based on the decision-making and planning algorithm to obtain a reference trajectory, a reference trajectory boundary, a reference speed and a reference speed boundary for driving or parking;

[0108] In step S104, a planning algorithm that meets the current scenario is called according to the decision-making scenario, and the reference trajectory and the reference trajectory boundary are subjected to trajectory planning by using the planning algorithm to generate an optimal trajectory for driving or parking, and the reference speed and the reference speed boundary are subjected to speed planning by using the planning algorithm to generate an optimal speed for driving or parking.

[0109] In step S105, a control algorithm that meets the current scenario is called according to the decision scenario, and the control algorithm is used to control the vehicle to perform a preset driving action or a preset parking action according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

[0110] It should be noted that the foregoing explanation of the platform embodiment of the autonomous vehicle also applies to the control method of the autonomous vehicle in this embodiment, and will not be elaborated here.

[0111] According to the control method of the autonomous vehicle provided by the embodiment of the present application, according to the vehicle system operation state and the decision scenario, a prediction algorithm suitable for the current scenario is called, and the prediction trajectory of the target around the vehicle is generated by combining the environmental information of the vehicle and the timing information of the target or obstacle. The optimal trajectory and speed of the vehicle for driving or parking are planned by using the vehicle body information, map positioning and other information of the vehicle, so as to realize the automatic driving or automatic parking of the vehicle. By adopting an integrated driving and parking planning and control architecture, the same software architecture is used for driving and parking. By calling the corresponding algorithm according to the scenario, the architecture can be shared and the modules can be shared, reducing the consumption of computing resources, reducing the number and cost of controllers, reducing the labor cost, and realizing that the platform algorithm can be applied to controllers and vehicle models with different computing powers through configuration files, realizing the platform development mode and improving the development efficiency. Thus, the problems that the related technology cannot realize the continuity of the daily travel scenario experience, and the driving and parking software architectures cannot be shared and the software scheduling efficiency is low are solved.

[0112] Figure 11 It is a block diagram of the vehicle provided by the embodiment of the present application.

[0113] As Figure 11 shown, the vehicle 20 includes: an autonomous driving system 600 and a software architecture platform 10 of the autonomous vehicle.

[0114] Among them, the preset driving function and the preset parking function of the autonomous driving system share a platform. The platform is used to obtain the environmental information, target information, operating state information, and decision-making scenarios around the vehicle; call the prediction algorithm that meets the current scenario according to the operating state information and decision-making scenarios, and perform trajectory prediction on the local environmental information and the target or obstacle time-series information based on the prediction algorithm to generate the predicted trajectory of the targets around the vehicle; call the decision-making algorithm that meets the current scenario according to the decision-making scenarios, and make decisions on the vehicle body information, map, positioning information, and predicted trajectory based on the decision-making and planning algorithm to obtain the reference trajectory, reference trajectory boundary, reference speed, and reference speed boundary for driving or parking; call the planning algorithm that meets the current scenario according to the decision-making scenarios, use the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate the optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate the optimal speed for driving or parking; call the control algorithm that meets the current scenario according to the decision-making scenarios, and use the control algorithm to control the vehicle to execute the preset driving actions or preset parking actions according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

[0115] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the control method of the autonomous driving vehicle as described above is implemented.

[0116] In the description of this specification, the descriptions with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, without conflict, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0117] In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0118] Any process or method description depicted in a flowchart or otherwise described herein may be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logical function or process. The scope of the preferred embodiments of the present application includes additional implementations where functions may be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0119] It should be understood that various parts of the present application may be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods may be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art may be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays, field programmable gate arrays, and the like.

[0120] Those of ordinary skill in the art of the present technology can understand that all or part of the steps carried out in implementing the methods of the above embodiments can be completed by instructing relevant hardware through a program. The said program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0121] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A software architecture platform for an autonomous vehicle, characterized in that, The vehicle is provided with a preset driving function and a preset parking function, and the platform is used to control the vehicle to implement the preset driving function and the preset parking function. Among them, the platform includes: An acquisition module, configured to acquire the local environment information sensed by the vehicle, the target or obstacle time series information, as well as the operating state information and decision-making scenario of the vehicle; A prediction module, configured to call a prediction algorithm that meets the current scenario according to the operating state information and decision-making scenario, and based on the prediction algorithm, perform trajectory prediction on the local environment information and the target or obstacle time series information to generate a predicted trajectory of the target around the vehicle; A decision-making module, configured to call a decision-making algorithm that meets the current scenario according to the decision-making scenario, and based on the decision-making algorithm, make decisions on the vehicle body information, map, positioning information, and the predicted trajectory of the vehicle to obtain a reference trajectory for driving or parking, a reference trajectory boundary, a reference speed, and a reference speed boundary; A planning module, configured to call a planning algorithm that meets the current scenario according to the decision-making scenario, use the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate an optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate an optimal speed for driving or parking; A control module, configured to call a control algorithm that meets the current scenario according to the decision-making scenario, and use the control algorithm to control the vehicle to execute a preset driving action or a preset parking action according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

2. The platform according to claim 1, wherein A scene recognition module is set to judge the current decision-making scenario according to the driving environment of the vehicle. The decision-making scenario includes one or more of a lane following scenario, an intersection scenario, a garage driving scenario, a garage parking scenario, and a U-turn scenario. These scenarios belong to driving decision-making scenarios and parking decision-making scenarios. The driving decision-making scenarios include a lane following scenario, an intersection scenario, a garage driving scenario, and a U-turn scenario. The parking decision-making scenario includes a garage parking scenario. Among them, If the vehicle meets the preset U-turn scenario condition, the vehicle enters the U-turn scenario, and then calls the prediction algorithm, decision-making algorithm, planning algorithm, and control algorithm required for the U-turn scenario; If the vehicle meets the preset intersection scenario condition, the vehicle enters the intersection scenario, and then calls the prediction algorithm, decision-making algorithm, planning algorithm, and control algorithm required for the intersection scenario; If the vehicle meets the preset garage driving scenario condition, the vehicle enters the garage scenario, and then calls the prediction algorithm, decision-making algorithm, planning algorithm, and control algorithm required for the garage driving scenario; If the vehicle meets the preset parking scenario condition, the vehicle enters the garage parking scenario, and then calls the prediction algorithm, decision-making algorithm, planning algorithm, and control algorithm required for the U-turn parking scenario.

3. The platform according to claim 1, wherein The decision-making module includes: A processing sub-module, which is used to determine a candidate lane set based on the actual type of the map, screen candidate lanes that meet preset conditions in the candidate lane set, determine a candidate coordinate system and boundaries according to the candidate lanes and a preset reference line, make a preset traffic rule decision according to the candidate coordinate system and boundaries to obtain a traffic rule decision result, and mark obstacles around the vehicle to obtain horizontal and vertical marking results of the obstacles; A decision-making sub-module, which is used to call a decision-making algorithm that meets the current scenario according to the decision-making scenario, use the decision-making algorithm to make a decision on the traffic rule decision result and the horizontal and vertical marking results of the obstacles, determine path boundary constraints and speed boundary constraints in the candidate coordinate system, and perform trajectory planning and trajectory feature extraction according to the path boundary constraints and speed boundary constraints to obtain a reference trajectory, a reference trajectory boundary, a reference speed, and a reference speed boundary for the vehicle to drive or park.

4. The platform according to claim 3, wherein Specifically, the processing sub-module is used for: If the map is a preset high-precision map, use a strategy of correcting the preset high-precision map and perception information to process the preprocessed data; If the map is a preset crowdsourced map, use a strategy of combining the preset crowdsourced map and perception information to process the preprocessed data, where the preset crowdsourced map is a road prior information graph generated by combining traffic elements sensed in real time multiple times.

5. The platform according to claim 4, characterized in that, The strategy of combining the preset crowdsourced map and perception information includes: Judge the lane information of the lane where the vehicle is currently located according to the preprocessed data; Perform positioning verification according to the lane information and the perception information to determine the lane identifier of the lane where the vehicle is currently located, search for a navigation reference center line using the lane identifier to obtain a reference path sequence, and determine the position relationship of the reference path center line points based on the reference path sequence; Perform mutual verification and fusion according to the perception information and the position relationship of the reference path center line points to obtain the left and right lane line relationships of the currently located lane, and perform mutual verification and fusion according to the perception information and the left and right lane line relationships to obtain lane attribute expansion information; Perform mutual verification and fusion according to the perception information, the lane attribute expansion information, and the reference line center line and boundary line point sequence, output a reference center line and boundary sequence, and determine a candidate lane set based on the reference center line and boundary sequence.

6. The platform according to claim 1, characterized in that, The prediction module includes: A data preprocessing sub-module, which is used to preprocess the environmental information and the target or obstacle time series information to obtain target information and preprocessed environmental information; A task scheduling sub-module, which is used to call a prediction algorithm that meets the current scenario according to the operation status information and the decision-making scenario; A prediction algorithm sub-module, which is used to perform destination prediction, intention prediction, and deep learning-based trajectory prediction on the target information and the preprocessed environmental information based on the prediction algorithm to obtain a predicted destination, a target intention, and a model-generated trajectory, obtain a rule-generated trajectory based on the predicted destination and the target intention, perform trajectory processing on the model-generated trajectory and the rule-generated trajectory to obtain a predicted trajectory of the target around the vehicle.

7. The platform according to claim 6, characterized in that, The task scheduling sub-module is specifically used for: If the vehicle is in a preset lane following scenario, call a preset driving prediction algorithm. Specifically, when the distance between the vehicle and the garage exit in the driving direction is less than a first threshold, the vehicle enters the preset lane following scenario; If the vehicle is in a preset garage driving scenario, call a preset garage driving scenario prediction algorithm. Specifically, when the distance between the vehicle and the garage entrance in the driving direction is less than a second threshold, the vehicle enters the preset garage driving scenario.

8. The platform according to claim 6, wherein The prediction module further includes: A data triggering sub-module, which is used to trigger and record preset scenario data and problem data, and upload the triggered data to the server to achieve data closed-loop.

9. The platform according to claim 1, wherein The planning module includes: A trajectory planning sub-module, which is used to plan an optimal path according to the reference trajectory, the reference trajectory boundary, the predicted trajectory, and the positioning information of the vehicle; A speed planning sub-module, which is used to plan an optimal speed according to the optimal path, the reference speed, and the reference speed boundary.

10. The platform according to claim 1, characterized in that, The control module adopts a control architecture combining feedforward and PID feedback, and the control architecture adopts a dual-loop PID control algorithm based on dynamically changing parameters.

11. A control method for an autonomous vehicle, characterized in that, The method is applied to a vehicle to control the vehicle to implement a preset driving function and a preset parking function. Specifically, the method includes the following steps: Obtain the local environment information, the target or obstacle time series information sensed by the vehicle, as well as the operating state information and decision-making scenario of the vehicle; According to the operating state information and decision-making scenario, call a prediction algorithm that meets the current scenario, and based on the prediction algorithm, perform trajectory prediction on the local environment information and the target or obstacle time series information to generate a predicted trajectory of the targets around the vehicle; According to the decision-making scenario, call a decision-making algorithm that meets the current scenario, and based on the decision-making planning algorithm, make decisions on the vehicle body information, map, positioning information, and the predicted trajectory of the vehicle to obtain a reference trajectory, a reference trajectory boundary, a reference speed, and a reference speed boundary for driving or parking; According to the decision-making scenario, call a planning algorithm that meets the current scenario, use the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate an optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate an optimal speed for driving or parking; According to the decision-making scenario, call a control algorithm that meets the current scenario, and use the control algorithm to control the vehicle to execute a preset driving action or a preset parking action according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

12. A vehicle, characterized in that, It includes: An autonomous driving system; A software architecture platform for an autonomous vehicle, wherein the preset driving function and the preset parking function of the autonomous driving system share the platform. The platform is used to obtain the environmental information, target information, operating status information, and decision-making scenarios around the vehicle; call a prediction algorithm that meets the current scenario according to the operating status information and decision-making scenarios, and perform trajectory prediction on the local environmental information and the target or obstacle time-series information based on the prediction algorithm to generate a predicted trajectory of the target around the vehicle; call a decision-making algorithm that meets the current scenario according to the decision-making scenario, and make a decision on the vehicle body information, map, positioning information, and the predicted trajectory based on the decision-making and planning algorithm to obtain a reference trajectory, reference trajectory boundary, reference speed, and reference speed boundary for driving or parking; call a planning algorithm that meets the current scenario according to the decision-making scenario, use the planning algorithm to perform trajectory planning on the reference trajectory and the reference trajectory boundary to generate an optimal trajectory for driving or parking, and use the planning algorithm to perform speed planning on the reference speed and the reference speed boundary to generate an optimal speed for driving or parking; call a control algorithm that meets the current scenario according to the decision-making scenario, and use the control algorithm to control the vehicle to execute a preset driving action or a preset parking action according to the optimal trajectory and the optimal speed, so as to realize the automatic driving and / or automatic parking of the vehicle.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor for implementing the control method of the autonomous vehicle as claimed in claim 11.

Images