Automatic driving algorithm verification method based on cloud control multi-vehicle experiment platform

Abstract

The invention relates to the technical field of vehicle-road cloud integration, in particular to an automatic driving algorithm verification method based on a cloud control multi-vehicle experiment platform, and the method comprises the steps: carrying out the deep analysis of the demands of a to-be-verified automatic driving algorithm, so as to adjust the content of the cloud control multi-vehicle experiment platform; by means of the real-time monitoring function of cloud management software, when the relative position and the posture state of a target vehicle do not meet the requirement in the requirement, the relative position and the posture state are adjusted in real time in a cloud remote control mode; sending an expected target vehicle track obtained by calculation and solution of a cloud algorithm in the cloud control multi-vehicle experiment platform to the target vehicle to obtain a verification result of the algorithm; and switching the to-be-tested comparison algorithm to obtain a comparison result of the to-be-verified automatic driving algorithm and the to-be-tested comparison algorithm. Therefore, the problems that the dynamic response of the actual target vehicle is difficult to reproduce truly, the communication function of the vehicle-road cloud integrated system is difficult to verify comprehensively, potential safety risks exist and the like in related technologies are solved.

Description

Technical Field

[0001] The present application relates to the field of vehicle-road-cloud integration technology, and in particular to an autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform. Background Art

[0002] With the rapid development of intelligent connected vehicle technology, the vehicle-road-cloud integrated system has become a research focus in academia and industry. This system uses advanced information and communication technologies to organically integrate the physical space and information space of people, vehicles, roads, and clouds to achieve collaborative perception, decision-making, and control. When developing collaborative control algorithms in vehicle-road-cloud integrated systems, various forms of verification tests are required. After completing simulation tests, software-in-the-loop tests, and hardware-in-the-loop tests, it is still necessary to test with real controlled objects to comprehensively evaluate realistic factors such as target vehicle dynamics and communication delays, thereby ensuring the effectiveness and reliability of the algorithm.

[0003] In related technologies, the cloud-based collaborative control algorithm can be verified through sandbox testing and real vehicle testing. The sandbox testing uses a miniature model car and a sandbox test bench for experiments, while the real vehicle testing can simulate real traffic scenarios.

[0004] However, the relevant technologies are limited by single road conditions and there is a large gap between them and the actual traffic environment. It is difficult to truly reproduce the dynamic response of the actual target vehicle. At the same time, sandbox testing usually cannot utilize necessary communication equipment such as OBU (On Board Unit) and RSU (Road Side Unit), making it difficult to fully verify the communication function of the vehicle-road-cloud integrated system. The actual vehicle test has high cost and low experimental efficiency, and there are potential safety risks. It is difficult to be widely used in the process of multiple algorithm iterations and optimization, and it is in urgent need of improvement. Summary of the invention

[0005] The present application provides an autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform to solve the problems in related technologies, such as sandbox testing is limited to a single road condition, which is quite different from the actual traffic environment, making it difficult to truly reproduce the dynamic response of the actual target vehicle, and making it difficult to fully verify the communication function of the vehicle-road-cloud integrated system. The actual vehicle testing is costly, inefficient, and has potential safety risks.

[0006] The first aspect of the present application provides an autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform. The method is applied to different autonomous driving verification algorithms such as cloud-controlled single-vehicle autonomous driving algorithm and cloud-controlled multi-vehicle collaborative control algorithm, and different road verification scenarios such as straight roads, curves, and intersections. The cloud management software is used to implement the switching of execution verification algorithms and verification scenarios, dynamically adjust the content of the cloud-controlled multi-vehicle experimental platform, adapt to different verification requirements, and accelerate the development and iteration of autonomous driving algorithms. The method includes the following steps: in the cloud-controlled algorithm adaptation and integration stage, an in-depth analysis is conducted on the requirements for the autonomous driving algorithm to be verified, so as to adjust the content of the cloud-controlled multi-vehicle experimental platform based on the requirements until the content meets the requirements, wherein the content includes at least one of a physical scaled model vehicle, the number of virtual target vehicles, and a deployment scenario; in the stage where the initial conditions of the cloud-controlled algorithm are achieved, the real-time monitoring function of the cloud management software is used to judge the requirements. Whether the relative position and posture state of the target vehicle in the cloud meets the requirements of the autonomous driving algorithm to be verified, if not, the relative position and posture state are adjusted in real time by remote control of the cloud until the relative position and posture state meet the requirements; in the dynamic execution and performance inspection stage of the cloud-controlled algorithm, the autonomous driving algorithm to be verified is executed, and the expected target vehicle trajectory calculated and solved by the cloud-based algorithm in the cloud-controlled multi-vehicle experimental platform is sent to the physical scaled model vehicle and the virtual target vehicle to obtain the verification result of the autonomous driving algorithm to be verified; in the cloud-controlled algorithm comparison test and performance evaluation stage, the cloud-controlled multi-vehicle experimental platform cloud is used to switch the autonomous driving algorithm to be verified to the comparison algorithm to be tested of the autonomous driving algorithm to be verified, and the above verification process is repeated, and based on the data collected in real time on the cloud, the comparison result of the autonomous driving algorithm to be verified and the comparison algorithm to be tested is obtained.

[0007] The second aspect of the present application provides an autonomous driving algorithm verification system based on a cloud-controlled multi-vehicle experimental platform, including: cloud-controlled multi-vehicle experimental platform management software, which is used to select the autonomous driving algorithm to be verified, and select the number of physical scaled model vehicles and virtual target vehicles according to the requirements of different algorithm verifications, and monitor the status information of the target vehicle in real time through a visual interface to obtain the control result of the autonomous driving algorithm to be verified; an autonomous driving algorithm running platform, which is used to run the autonomous driving algorithm to be verified, and generate a reference trajectory corresponding to the target vehicle according to the status information of the target vehicle; and an autonomous driving algorithm execution target vehicle, which is used to track the reference trajectory issued by the cloud and respond to cloud control instructions in real time.

[0008] The third aspect of the present application provides a cloud-controlled multi-vehicle experimental platform, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the autonomous driving algorithm verification method based on the cloud-controlled multi-vehicle experimental platform as described in the above embodiment.

[0009] The fourth aspect embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform.

[0010] The fifth aspect embodiment of the present application provides a computer program product, including a computer program, which, when executed, implements the above-mentioned autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform.

[0011] In the cloud control algorithm adaptation and integration stage, the embodiment of the present application can conduct an in-depth analysis of the requirements of the autonomous driving algorithm to be verified, so as to adjust the content of the cloud control multi-vehicle experimental platform until the content meets the requirements. In the cloud control algorithm initial condition achievement stage, the real-time monitoring function of the cloud management software is used. When the relative position and posture state of the target vehicle in the requirements do not meet the requirements of the autonomous driving algorithm to be verified, the relative position and posture state are adjusted in real time through cloud remote control until the relative position and posture state meet the requirements. In the cloud control algorithm dynamic execution and performance test stage, the autonomous driving algorithm to be verified is executed to obtain the verification result of the autonomous driving algorithm to be verified. In the cloud control algorithm comparison test and performance evaluation stage, the cloud control multi-vehicle experimental platform is used to switch the autonomous driving algorithm to be verified to the comparison algorithm to be tested of the autonomous driving algorithm to be verified, so as to obtain the comparison result of the autonomous driving algorithm to be verified and the comparison algorithm to be tested, so as to realize the rapid deployment, real-time monitoring and remote control of the algorithm, and improve the flexibility and security of verification through cloud control and data collection, and provide data support for algorithm optimization. This solves the problems in related technologies, such as sandbox testing being limited to single road conditions, having a large gap with the actual traffic environment, making it difficult to truly reproduce the dynamic response of the actual target vehicle, and making it difficult to fully verify the communication function of the vehicle-road-cloud integrated system. The actual vehicle testing is costly, inefficient, and has potential safety risks.

[0012] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0014] Figure 1 A block diagram of a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application;

[0015] Figure 2 A flowchart of an autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application;

[0016] Figure 3 A flowchart of an information collection unit determining a controlled state of a target vehicle according to an embodiment of the present application;

[0017] Figure 4 A schematic diagram of the structure of the logic for implementing the autonomous obstacle avoidance unit provided according to one embodiment of the present application;

[0018] Figure 5 A block diagram of an autonomous driving algorithm verification system based on a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application;

[0019] Figure 6 This is a schematic diagram of the structure of a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0021] The following describes the autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform according to an embodiment of the present application with reference to the accompanying drawings. The sandbox test mentioned in the above background technology is limited to a single road condition, which is quite different from the actual traffic environment. It is difficult to truly reproduce the dynamic response of the actual target vehicle, and it is difficult to fully verify the communication function of the vehicle-road-cloud integrated system. The actual vehicle test has high cost and low efficiency, and there are potential safety risks. The present application provides an autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform. In this method, in the cloud-controlled algorithm adaptation and integration stage, the requirements for the autonomous driving algorithm to be verified can be deeply analyzed to adjust the content of the cloud-controlled multi-vehicle experimental platform until the content meets the requirements. In the stage where the initial conditions of the cloud-controlled algorithm are achieved, the real-time monitoring function of the cloud management software is used to check if the relative position and posture state of the target vehicle do not meet the requirements. When the demand for the autonomous driving algorithm to be verified is met, the relative position and posture state are adjusted in real time through remote control in the cloud until the relative position and posture state meet the demand. In the cloud control algorithm dynamic execution and performance test stage, the autonomous driving algorithm to be verified is executed to obtain the verification result of the autonomous driving algorithm to be verified. In the cloud control algorithm comparison test and performance evaluation stage, the cloud control multi-vehicle experimental platform is used to switch the autonomous driving algorithm to be verified to the comparison algorithm to be tested of the autonomous driving algorithm to be verified, and then the comparison result of the autonomous driving algorithm to be verified and the comparison algorithm to be tested is obtained, thereby realizing the rapid deployment, real-time monitoring and remote control of the algorithm, and through cloud control and data collection, the flexibility and security of verification are improved, and data support is provided for algorithm optimization. In this way, the problems in the related technologies that the sandbox test is limited to a single road condition and there is a large gap with the actual traffic environment, it is difficult to truly reproduce the dynamic response of the actual target vehicle, it is difficult to fully verify the communication function of the vehicle-road-cloud integrated system, the actual vehicle test is costly and inefficient, and there are potential safety risks.

[0022] Before introducing the autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform proposed in an embodiment of the present application, the cloud-controlled multi-vehicle experimental platform involved in the embodiment of the present application is first introduced.

[0023] Specifically, Figure 1 A block diagram of a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application.

[0024] like Figure 1 As shown, the cloud-controlled multi-vehicle experimental platform 10 includes: a cloud end 110 , a road end 120 and a vehicle end 130 .

[0025] The cloud 110 may include but is not limited to cloud management software 111, cloud server 112 and driving simulator 113, etc., and this application does not impose any specific restrictions.

[0026] Furthermore, in the embodiment of the present application, the cloud 110 is mainly composed of a cluster of multiple high-performance cloud servers 112. Each cloud server 112 is not only equipped with an XX processor (such as 2.0GHz / 10 cores / 6.4GT / 25MB cache, etc., which is not specifically limited in this application), and 4 16GB DDR4 (Double Data Rate Fourth Generation Synchronous Dynamic Random Access Memory) memories with a total capacity of 64GB, but also equipped with a 300GB hot-swappable SAS (Serial Attached Small Computer System Interface) hard drive (10,000 rpm), 5 2.5-inch hard drive slots, XX company's eight-channel SAS RAID (Redundant Array of Independent Disks) card (RS0820P, 2GB cache), redundant power supply, external DVD (Digital Versatile Disc) drive and 4 Gigabit network cards. In addition, in the embodiment of the present application, the cloud 110 is also provided with other standard configurations, which may include but are not limited to guide rails, a display card and a 24GB XXGPU (Graphics Processing Unit).

[0027] In addition, it should be noted that in the embodiment of the present application, Kafka and MQTT (Message Queuing Telemetry Transport) services are deployed in the cloud server 112 cluster to realize message transmission based on the publish-subscribe model; the cloud algorithm runs on the Storm server, which can process the target vehicle status information in real time and efficiently issue control instructions. At the same time, the cloud 110 has also developed cloud management software based on the B / S (Browser / Server) architecture, which is responsible for managing cloud algorithms and target vehicles, thereby simplifying the algorithm verification process. In addition, in order to realize remote control of experimental target vehicles, the cloud 110 is also equipped with a driving simulator 113.

[0028] The road end 120 may include, but is not limited to, an RSU 121 and a roadside sensing device 122 , such as a high-definition camera 122 , etc., and this application does not impose any specific limitations.

[0029] Furthermore, in the embodiment of the present application, RSU121 adopts the intelligent roadside system of XX company, supporting network communication between the roadside and the cloud 110, and between the road end 120 and the vehicle end 130; the high-definition camera 122 adopts the high-definition network camera of XX company, which is used for real-time monitoring of the experimental site, and this application does not make any specific restrictions.

[0030] The vehicle end 130 may include, but is not limited to, an information collection unit 131 , an autonomous obstacle avoidance unit 132 , and a target vehicle control unit 133 .

[0031] Furthermore, in the embodiment of the present application, the vehicle end 130 is composed of three small vehicles, each of which has a size of 1435mm×840mm×450mm, and is equipped with a wire-controlled chassis of XX Company, and the chassis feedback information is read and controlled through a reserved CAN (Controller Area Network) bus interface. Among them, the chassis integrates the positioning equipment of XX Company to provide high-precision RTK (Real-Time Kinematic Positioning Information) positioning information, and the communication equipment uses the intelligent vehicle-mounted terminal of XX Company to realize real-time communication between the vehicle and the road, and is equipped with a laser radar of XX Company for obstacle perception.

[0032] In addition, it should be noted that the cloud-controlled multi-vehicle experimental platform of the embodiment of the present application has three core functions, namely, target vehicle status monitoring, algorithm vehicle control and remote control vehicle control.

[0033] Among them, the target vehicle status monitoring function is intended to monitor the status of the target vehicle in real time through the cloud management software 111. The information collection unit 131 obtains the status information of the target vehicle, such as position, speed and acceleration, through the CAN bus, and determines the controlled state of the target vehicle by checking the startup status of each node and the reception of cloud commands. Subsequently, the status information is packaged and broadcast to RSU121 via C-V2X (Cellular-Vehicle to Everything), and RSU121 then sends the status information of all target vehicles to the cloud 110 via the MQTT protocol. The road-side high-definition camera 122 captures the screen of the experimental site in real time and transmits the video to the cloud 110 via the MQTT protocol. After the cloud 110 receives the target vehicle status information and video screen, these data will be displayed in real time in the cloud management software 111 interface to achieve real-time monitoring of the target vehicle status.

[0034] The algorithm control vehicle function is designed to select the experimental target vehicle and run the autonomous driving algorithm to be verified through the cloud management software 111. The cloud-based autonomous driving algorithm to be verified is deployed on the Nimbus node of the Storm server, which obtains the target vehicle status information by subscribing to messages in the Kafka and MQTT cloud servers 112. In order to simplify the verification process and improve efficiency, the cloud management software 111 has the algorithm selection and target vehicle matching functions. After receiving the target vehicle status information, the cloud management software 111 sends the status information of the selected experimental target vehicle to the specified algorithm, and realizes the automatic matching of the target vehicle ID (IDentity) in the specified algorithm with the experimental target vehicle ID without modifying the code. After the autonomous driving algorithm to be verified obtains the experimental target vehicle status information, it generates the corresponding expected trajectory through calculation and processing, and sends the expected trajectory to RSU121 through Kafka and MQTT cloud servers 112. After RSU121 receives and processes the information, it broadcasts the target vehicle identifier and expected trajectory to each target vehicle through C-V2X. The target vehicle identifies its expected trajectory based on the target vehicle ID, and after receiving the trajectory, the onboard controller uses PID (Proportional Integral Derivative) and MPC (Model Predictive Control) algorithms for longitudinal and lateral control, thereby achieving accurate trajectory tracking. At this point, the complete process of the autonomous driving algorithm to be verified has been realized.

[0035] The remote control function is used to remotely control the target vehicle to the designated experimental starting point before the target vehicle is controlled by the autonomous driving algorithm to be verified, and reset the target vehicle to repeat the experiment after the experiment is over or when the algorithm runs incorrectly and causes the target vehicle to deviate from the route. Any experimental target vehicle can be selected for remote control through the cloud management software 111. The remote control command is generated by the driving simulator 113 and sent to each target vehicle through the same communication link as the algorithm control vehicle. The chassis of the target vehicle responds to the received control command to achieve remote control operation.

[0036] In addition, in order to deal with the potential safety hazards that may occur during the verification of the autonomous driving algorithm to be verified, whether there is a problem with the global reference trajectory issued by the cloud 110 or the trajectory encounters an obstacle in the actual scene, the experimental platform introduces an emergency brake mechanism in the cloud management software 111, and adds an autonomous obstacle avoidance unit 132 on the vehicle side 130 to achieve autonomous obstacle avoidance and ensure the safety of algorithm verification. When the cloud management software 111 controls the vehicle by algorithm or remotely controls the experimental target vehicle, it can perform emergency brake operations on any or all target vehicles at any time to prevent potential risks. At the same time, the autonomous obstacle avoidance unit 132 senses obstacles through laser radar and processes the point cloud to obtain the center position and size of the obstacle. Subsequently, the system performs local optimal path planning through CILQR, combined with constraints such as actuator constraints and obstacle avoidance constraints, to ensure that the driving trajectory of the experimental target vehicle is as close to the global reference trajectory of the cloud 110 as possible under the premise of safe obstacle avoidance.

[0037] Specifically, Figure 2 The present invention provides a flowchart of an autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform according to an embodiment of the present application.

[0038] like Figure 2 As shown, the autonomous driving algorithm verification method based on the cloud-controlled multi-vehicle experimental platform is applied to different autonomous driving verification algorithms such as cloud-controlled single-vehicle autonomous driving algorithm and cloud-controlled multi-vehicle cooperative control algorithm, and different road verification scenarios such as straight roads, curves, and intersections. The cloud management software is used to switch between the execution verification algorithm and the verification scenario, dynamically adjust the content of the cloud-controlled multi-vehicle experimental platform, adapt to different verification requirements, and accelerate the development and iteration of the autonomous driving algorithm. The method includes the following steps:

[0039] In step S201, during the cloud-controlled algorithm adaptation and integration stage, an in-depth analysis is conducted on the requirements for the autonomous driving algorithm to be verified, so as to adjust the content of the cloud-controlled multi-vehicle experimental platform based on the requirements until the content meets the requirements, wherein the content includes at least one of a physical scaled model vehicle, a number of virtual target vehicles, and a deployment scenario.

[0040] It can be understood that the autonomous driving algorithm to be verified in the embodiments of the present application may include but is not limited to a cloud-controlled single-vehicle autonomous driving algorithm, a cloud-controlled multi-vehicle collaborative control algorithm, etc. This application does not make any specific restrictions.

[0041] Furthermore, in an embodiment of the present application, the requirements may include data of different working scenarios such as straight roads, curves, and intersections, and the switching between the execution of the autonomous driving algorithm to be verified and the working scenario can be achieved through cloud management software, and the content of the cloud-controlled multi-vehicle experimental platform can be dynamically adjusted to adapt to different verification requirements, thereby accelerating the development and iteration of the autonomous driving algorithm to be verified; in addition, the requirements may also include information about the surrounding areas of the target vehicle's laser radar, information about the wire-controlled chassis, RTK positioning information of the target vehicle, etc., which are not specifically limited in this application.

[0042] In addition, the embodiments of the present application can conduct an in-depth analysis of the requirements for the autonomous driving algorithm to be verified, and adjust the content of the cloud-controlled multi-vehicle experimental platform as needed, which may include but is not limited to adjusting the number of physical scaled model vehicles, virtual target vehicles and deployment scenarios, etc. Through the developed cloud management software, the rapid deployment of the autonomous driving algorithm to be verified can be achieved, ensuring the efficiency and accuracy of the experimental preparation work.

[0043] It should be noted that in the embodiment of the present application, the cloud management software can select the autonomous driving algorithm to be verified, and select the number of controlled physical scaled model target vehicles and virtual target vehicles according to the requirements of different algorithms; monitor the status information of the controlled target vehicles in real time through the visual interface, and display the control effect of the autonomous driving algorithm to be verified.

[0044] In some embodiments, in combination Figure 1 In the embodiment of the present application, the vehicle side of the cloud-controlled multi-vehicle experimental platform is started, and the information collection unit starts working to obtain the requirements of the autonomous driving algorithm to be verified through Ethernet, CAN line, and serial port, such as laser radar surrounding information, wire-controlled chassis information, and GPS positioning information, etc. This application does not make specific restrictions.

[0045] Furthermore, in the embodiment of the present application, the laser radar obtains point cloud data information, uses Ethernet for data transmission, and then obtains the center position and size information of the obstacle through point cloud processing; the speed, acceleration, driving mode and other information in the wire-controlled chassis are sent to the CAN bus, and the information collection unit obtains the information of the wire-controlled chassis on the CAN bus through the CAN to Ethernet protocol; R60s obtains RTK positioning information and transmits data through the serial port.

[0046] In addition, the information collection unit of the embodiment of the present application can also determine the controlled state of the target vehicle, and its process is as follows: Figure 3 As shown, the main steps are:

[0047] Step S301: Start the obu_send node.

[0048] Step S302: Determine whether all nodes have been started.

[0049] In this embodiment of the present application, when all nodes are started, step S303 is executed; otherwise, step S305 is executed.

[0050] Step S303: Whether the cloud sends an instruction.

[0051] In which, in the embodiment of the present application, when an instruction is issued, step S304 is executed; otherwise, step S306 is executed.

[0052] Step S304: Determine the type of instruction sent by the cloud.

[0053] In this embodiment of the present application, when the instruction type is a reference trajectory, step S307 is executed; when the instruction type is a control instruction, step S308 is executed.

[0054] Step S305: Update the controlled state of the target vehicle to a not-ready state.

[0055] Step S306: Update the controlled state of the target vehicle to the standby state.

[0056] Step S307: updating the controlled state of the target vehicle to a remote control state.

[0057] Step S308: Update the controlled state of the target vehicle to the algorithm-controlled state.

[0058] In step S202, when the initial conditions of the cloud control algorithm are met, the real-time monitoring function of the cloud management software is used to determine whether the relative position and posture state of the target vehicle in the demand meets the requirements of the autonomous driving algorithm to be verified. If not, the relative position and posture state are adjusted in real time through cloud remote control until the relative position and posture state meet the requirements.

[0059] As a possible implementation method, the embodiment of the present application can use the real-time monitoring function of the cloud management software to determine whether the relative position and posture state of the target vehicle in the demand meets the requirements of the autonomous driving algorithm to be verified. If not, the state of the target vehicle is adjusted in real time through cloud remote control until the relative position and posture state meet the requirements, thereby reducing the necessity of personnel presence and improving the flexibility of the verification process.

[0060] Illustratively, after the information collection unit of the embodiment of the present application collects the requirements of all target vehicles of the autonomous driving algorithm to be verified, such as target vehicle status information and surrounding environment information, the target vehicle status information can be packaged into a JSON (JavaScript Object Notation) data packet through C-V2X communication and sent to the road-side RSU, and the surrounding environment information will be sent to the autonomous obstacle avoidance unit in the form of a Topic through ROS.

[0061] Furthermore, after receiving the target vehicle status information sent by the vehicle side, the road-side RSU of the embodiment of the present application will determine to which target vehicle the target vehicle status information belongs, and determine whether the corresponding initial state meets certain initial conditions, and then send the target vehicle status information to the corresponding MQTT Topic; the road-side high-definition camera will collect real-time images of the experimental site in real time, and then transmit it to the cloud through the MQTT protocol.

[0062] Then continue until the relative position and posture state meet the requirements.

[0063] In step S203, during the dynamic execution and performance testing phase of the cloud-controlled algorithm, the autonomous driving algorithm to be verified is executed, and the desired target vehicle trajectory obtained by calculating and solving the cloud-based algorithm in the cloud-controlled multi-vehicle experimental platform is sent to the physical scaled model vehicle and the virtual target vehicle to obtain the verification result of the autonomous driving algorithm to be verified.

[0064] During the actual execution process, the embodiment of the present application can execute the autonomous driving algorithm to be verified when the relative position and posture status meet the requirements, generate the corresponding expected target vehicle trajectory based on the status information of the target vehicle, track the expected target vehicle trajectory sent to the cloud, and respond to cloud control instructions in real time.

[0065] In addition, it should be noted that the embodiment of the present application executes the autonomous driving algorithm to be verified, sends the expected target vehicle trajectory obtained by calculating and solving the cloud-based algorithm to the physical scaled model vehicle and the virtual target vehicle, and verifies the actual effect of the algorithm. In the embodiment of the present application, the vehicle-side autonomous obstacle avoidance unit of the cloud-controlled multi-vehicle experimental platform has an autonomous obstacle avoidance function. Combined with the real-time monitoring function of the cloud, it can take remote takeover measures when abnormal situations are found to ensure the safety of the verification process.

[0066] In some embodiments, the embodiments of the present application can determine the road-side state information and road-side surrounding environment information of the cloud-controlled multi-vehicle experimental platform based on the initial vehicle-side state information and initial vehicle-side surrounding environment information of all target vehicles, and then obtain the corresponding final vehicle-side state information and final vehicle-side surrounding environment information, thereby calculating the reference target vehicle trajectory of all target vehicles and generating the expected target vehicle trajectory of all target vehicles.

[0067] Exemplarily, the cloud management software of the embodiment of the present application obtains the initial vehicle-side status information and initial vehicle-side surrounding environment information of the target vehicle by subscribing to the relevant MQTT Topic, which may include but is not limited to the controlled state, position information, speed information, etc., and displays it on the software interface after processing.

[0068] Furthermore, in an embodiment of the present application, the cloud management software can send the initial vehicle-side state information and initial vehicle-side surrounding environment information of the target vehicle to the autonomous driving algorithm to be verified as algorithm input, and can realize the automatic matching of the target vehicle ID in the algorithm with the experimental target vehicle ID without modifying the code. After obtaining the initial vehicle-side state information and initial vehicle-side surrounding environment information of the experimental target vehicle, the autonomous driving algorithm to be verified generates the corresponding reference target vehicle trajectory through calculation and processing, and sends the algorithm output trajectory to the RSU through the Kafka and MQTT cloud servers.

[0069] Furthermore, in an embodiment of the present application, the cloud management software can select any target vehicle to enter the remote control state, and send the remote control instructions generated by the driving simulator to the RSU through the Kafka and MQTT cloud servers.

[0070] Furthermore, in an embodiment of the present application, the cloud management software can select any target vehicle or all target vehicles to perform an emergency brake operation when the vehicle is controlled by algorithm or remote control, and send the emergency brake command to the RSU through the Kafka and MQTT cloud server.

[0071] Furthermore, in the embodiment of the present application, the road-side RSU subscribes to the Topic of MQTT, receives the reference target vehicle trajectory and remote control instructions sent from the cloud, and sends the information to the experimental target vehicle.

[0072] Among them, in the embodiment of the present application, if the information collection unit on the vehicle side receives the reference target vehicle trajectory sent from the cloud, the autonomous obstacle avoidance unit will perform local path planning based on the final vehicle side status information and the final vehicle side surrounding environment information, and plan a safe local obstacle avoidance trajectory on the basis of ensuring that it is as close as possible to the reference target vehicle trajectory on the cloud, thereby ensuring the safety of the sent trajectory. After processing, the expected target vehicle trajectory will be sent to the target vehicle control unit in the form of a Topic through ROS (Robot Operating System), and the target vehicle control unit will use PID and MPC control to accurately track the trajectory; if the vehicle-side information collection unit receives a remote control command sent from the cloud, it will send the control command directly to the CAN bus to achieve direct control of the wire-controlled chassis.

[0073] In some embodiments, when the vehicle-side information collection unit of the embodiment of the present application receives the reference target vehicle trajectory sent from the cloud, the vehicle-side autonomous obstacle avoidance unit will perform local optimal path planning using the CILQR planning model based on the road-side state information, road-side surrounding environment information, final vehicle-side state information and final vehicle-side surrounding environment information obtained by the information collection unit, combined with actuator constraints and obstacle avoidance constraints, to obtain the reference target vehicle trajectory.

[0074] For example, the implementation logic of the autonomous obstacle avoidance unit in the embodiment of the present application is as follows: Figure 4 As shown, the CILQR planning model adds constraints on the basis of the original ILQR algorithm. The expression of the established CILQR planning model can be, but is not limited to,:

[0075]

[0076] stx k+1 =f k (x k ,u k ),k=0,1,…,N-1,

[0077] x0=x start ,

[0078] g(x,u)<0,

[0079] h(x,u)=0,

[0080] Among them, x k ,u k are the state vector and input control vector at time k, The state vector at time k includes the lateral and longitudinal positions, speeds, and heading angles at time k. The control vector at time k includes the acceleration and angle at time k. N is the prediction time domain. φ(x N ) is the terminal loss function, L k (x k ,u k ) is the loss function at the k-th moment, and the two together constitute the objective function that the system needs to optimize, f k (x k ,u k ) is the dynamic equation of the system, x start is the initial state constraint, and the last two equations are the inequality and equality constraints respectively.

[0081] Further, in the embodiment of the present application, the objective function to be optimized can be decoupled into a control loss function and a state loss function, the control loss function includes an acceleration loss function and an angle loss function, the state loss function includes a reference trajectory tracking loss function and a reference speed tracking loss function, and then the total objective function is obtained, and its calculation formula can be but is not limited to:

[0082]

[0083] Among them, w acc is the acceleration weight, w stetr is the corner weight, w vel is the speed weight, is the trajectory tracking weight matrix, v r is the expected speed, x k r is the expected state vector.

[0084] Further, in the embodiment of the present application, after obtaining the objective function to be optimized, the state vector may be iterated according to the target vehicle kinematic model, wherein the expression of the target vehicle kinematic model may be, but is not limited to,:

[0085] v1=v0+aTr r ,

[0086]

[0087] Among them, T r is the sampling time. It should be noted that the above equation is based on the assumption that the rotation angle remains unchanged during the sampling time. is the distance the target vehicle moves within one sampling time, is the curvature, and L is the wheelbase of the target vehicle.

[0088] Furthermore, in the embodiments of the present application, actuator constraints and obstacle avoidance constraints may include, but are not limited to, acceleration constraints, angle constraints, obstacle avoidance constraints, etc., and the present application does not impose any specific limitations.

[0089] The acceleration constraint is limited by the throttle opening and the brake cylinder pressure, and its expression can be but is not limited to:

[0090]

[0091] Among them, a high >0 provides acceleration for maximum throttle opening, a low <0 provides acceleration for maximum wheel cylinder pressure.

[0092] The expression of the rotation constraint can be, but is not limited to:

[0093]

[0094] in, is the maximum turning angle of the target vehicle.

[0095] The obstacle avoidance constraint models the obstacle as a rectangle and limits the distance from the vehicle to the rectangle to ensure obstacle avoidance safety. Its expression can be, but is not limited to,:

[0096]

[0097] Among them, d(x k ,y) represents point x k The Euclidean distance from point y.

[0098] Furthermore, after obtaining the reference target vehicle trajectory, the embodiment of the present application needs to post-process the trajectory. According to the target vehicle position obtained by the information collection unit, the nearest trajectory point is found, and the planned trajectory after the nearest trajectory point is interpolated with a fifth-order polynomial equidistant interpolation, and then sent to the target vehicle control unit to ensure the subsequent trajectory tracking effect.

[0099] Furthermore, the embodiment of the present application can obtain the initial reference target vehicle trajectory of all target vehicles based on the CILQR planning model, and based on the initial reference target vehicle trajectory, adjust the initial longitudinal deviation and the initial lateral deviation of the initial reference target vehicle trajectory to obtain the adjusted longitudinal deviation and the adjusted lateral deviation, and then obtain the reference target vehicle trajectory.

[0100] Exemplarily, the vehicle-side target vehicle control unit of the embodiment of the present application includes longitudinal control and lateral control of the target vehicle, wherein the longitudinal control uses PID control to track the desired speed, and the lateral control uses MPC control to output the front wheel steering angle to track the desired trajectory.

[0101] Specifically, in the embodiment of the present application, the target vehicle longitudinal speed is controlled by using the driving torque and the brake wheel cylinder pressure. Therefore, based on the deviation v between the longitudinal vehicle speed and the expected vehicle speed, err Construct acceleration and deceleration PID controllers, input the driving torque and braking pressure to the car respectively, and control the longitudinal speed of the car. The expression can be but not limited to:

[0102]

[0103] The target vehicle lateral MPC control is based on the target vehicle two-degree-of-freedom bicycle model, and the state variable state = [xy ψ ve trk e ψ ], control quantity u t =[δ t ,a t ]Use the current state to predict the next state. The model expression can be but is not limited to:

[0104]

[0105] It should be noted that in the embodiment of the present application, the core of the model predictive control is the design and solution of the objective function. By establishing a prediction model, the optimal control strategy is solved using an optimization algorithm. The core goal of trajectory tracking control is to minimize the deviation between the actual trajectory and the desired trajectory, that is, to minimize the lateral deviation e of the car at each moment. trk The heading angle deviation e ψAt the same time, in order to reduce impact loss and improve ride comfort, it is necessary to limit the rate of change of the control amount, that is, the rate of change of the front wheel angle Δδ t , acceleration change rate Δa t It should be as small as possible. At the same time, the longitudinal velocity error, front wheel angle and acceleration are added with soft constraints, and the weight coefficients of the corresponding items are adjusted to achieve better trajectory tracking effect. The objective function expression of MPC lateral control can be but is not limited to:

[0106]

[0107] Furthermore, the embodiment of the present application can transform the above trajectory tracking problem into a quadratic programming problem, and its expression can be but not limited to:

[0108]

[0109] The embodiment of the present application can use the nonlinear optimization problem solver Ipopt to solve the above-mentioned constrained quadratic programming problem to obtain the optimal control quantity in the set control time domain. The first of the obtained control quantity sequence is applied to the controlled system to obtain the actual optimal control quantity, that is, the front wheel steering angle.

[0110] In addition, the embodiments of the present application take into account that only using MPC to control the steering wheel angle is difficult to meet various road conditions and is not universal. At the same time, because only closed-loop feedback control is used, the car may not turn in time at the straight turn, resulting in large tracking errors and lateral error overshoots, affecting the trajectory tracking effect. Taking into account the above limitations, the angle is corrected and compensated by pre-aiming curvature feedforward and switching MPC weight parameters, thereby reducing the lateral tracking error and improving the trajectory tracking performance. Among them, the curvature feedforward angle compensation expression can be but is not limited to:

[0111] w feedforward =arctan(2.75κ),

[0112] Where κ is the curvature of the tracked trajectory.

[0113] In step S204, during the cloud-controlled algorithm comparison test and performance evaluation stage, the cloud-controlled multi-vehicle experimental platform is used to switch the autonomous driving algorithm to be verified to the comparison algorithm to be tested of the autonomous driving algorithm to be verified, and the above verification process is repeated. Based on the data collected in real time on the cloud, a comparison result between the autonomous driving algorithm to be verified and the comparison algorithm to be tested is obtained.

[0114] It will be understood by those skilled in the art that the embodiments of the present application can switch the comparison algorithm to be tested in the cloud, repeat the above-mentioned verification process, compare the experimental results of the autonomous driving algorithm to be verified and the comparison algorithm based on the data collected in real time on the cloud, evaluate the performance difference, provide a basis for further optimization of the algorithm, and then obtain the comparison result of the autonomous driving algorithm to be verified and the comparison algorithm to be tested.

[0115] According to the automatic driving algorithm verification method based on the cloud-controlled multi-vehicle experimental platform proposed in the embodiment of the present application, in the cloud-controlled algorithm adaptation and integration stage, the requirements of the automatic driving algorithm to be verified can be deeply analyzed to adjust the content of the cloud-controlled multi-vehicle experimental platform until the content meets the requirements. In the initial condition achievement stage of the cloud-controlled algorithm, the real-time monitoring function of the cloud management software is used. When the relative position and posture state of the target vehicle in the requirements do not meet the requirements of the automatic driving algorithm to be verified, the relative position and posture state are adjusted in real time through cloud remote control until the relative position and posture state meet the requirements. In the cloud-controlled algorithm dynamic execution and performance test stage, the automatic driving algorithm to be verified is executed to obtain the verification result of the automatic driving algorithm to be verified. In the cloud-controlled algorithm comparison test and performance evaluation stage, the cloud-controlled multi-vehicle experimental platform is used to switch the automatic driving algorithm to be verified to the comparison algorithm to be tested of the automatic driving algorithm to be verified, and then the comparison result of the automatic driving algorithm to be verified and the comparison algorithm to be tested is obtained, thereby realizing rapid deployment, real-time monitoring and remote control of the algorithm, and improving the flexibility and security of verification through cloud control and data collection, providing data support for algorithm optimization. This solves the problems in related technologies, such as sandbox testing being limited to single road conditions, having a large gap with the actual traffic environment, making it difficult to truly reproduce the dynamic response of the actual target vehicle, and making it difficult to fully verify the communication function of the vehicle-road-cloud integrated system. The actual vehicle testing is costly, inefficient, and has potential safety risks.

[0116] Secondly, the autonomous driving algorithm verification system based on the cloud-controlled multi-vehicle experimental platform proposed in accordance with the embodiment of the present application is described with reference to the accompanying drawings.

[0117] Figure 5 A block diagram of an autonomous driving algorithm verification system based on a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application.

[0118] like Figure 5 As shown, the autonomous driving algorithm verification system 50 based on the cloud-controlled multi-vehicle experimental platform includes: a cloud-controlled multi-vehicle experimental platform management software 100, an autonomous driving algorithm running platform 200 and an autonomous driving algorithm execution target vehicle 300.

[0119] Among them, the cloud-controlled multi-vehicle experimental platform management software 100 is used to select the autonomous driving algorithm to be verified, and select the number of physical scaled model vehicles and virtual target vehicles according to the requirements of different algorithm verifications, and monitor the status information of the target vehicle in real time through a visual interface to obtain the control results of the autonomous driving algorithm to be verified.

[0120] The autonomous driving algorithm operation platform 200 is used to run the autonomous driving algorithm to be verified and generate a reference trajectory corresponding to the target vehicle according to the state information of the target vehicle.

[0121] The autonomous driving algorithm executes the target vehicle 300 to track the reference trajectory sent from the cloud and respond to the cloud control instructions in real time.

[0122] It should be noted that the aforementioned explanation of the embodiment of the autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform is also applicable to the autonomous driving algorithm verification system based on a cloud-controlled multi-vehicle experimental platform of this embodiment, and will not be repeated here.

[0123] According to the autonomous driving algorithm verification system based on the cloud-controlled multi-vehicle experimental platform proposed in the embodiment of the present application, in the cloud-controlled algorithm adaptation and integration stage, the requirements of the autonomous driving algorithm to be verified can be deeply analyzed to adjust the content of the cloud-controlled multi-vehicle experimental platform until the content meets the requirements. In the initial condition achievement stage of the cloud-controlled algorithm, the real-time monitoring function of the cloud management software is used. When the relative position and posture state of the target vehicle in the requirements do not meet the requirements of the autonomous driving algorithm to be verified, the relative position and posture state are adjusted in real time through cloud remote control until the relative position and posture state meet the requirements. In the cloud-controlled algorithm dynamic execution and performance test stage, the autonomous driving algorithm to be verified is executed to obtain the verification result of the autonomous driving algorithm to be verified. In the cloud-controlled algorithm comparison test and performance evaluation stage, the cloud-controlled multi-vehicle experimental platform is used to switch the autonomous driving algorithm to be verified to the comparison algorithm to be tested of the autonomous driving algorithm to be verified, and then the comparison result of the autonomous driving algorithm to be verified and the comparison algorithm to be tested is obtained, thereby realizing rapid deployment, real-time monitoring and remote control of the algorithm, and improving the flexibility and security of verification through cloud control and data collection, providing data support for algorithm optimization. This solves the problems in related technologies, such as sandbox testing being limited to single road conditions, having a large gap with the actual traffic environment, making it difficult to truly reproduce the dynamic response of the actual target vehicle, and making it difficult to fully verify the communication function of the vehicle-road-cloud integrated system. The actual vehicle testing is costly, inefficient, and has potential safety risks.

[0124] Figure 6 This is a schematic diagram of the structure of a cloud-controlled multi-vehicle experimental platform provided according to an embodiment of the present application. The cloud-controlled multi-vehicle experimental platform may include:

[0125] A memory 601 , a processor 602 , and a computer program stored in the memory 601 and executable on the processor 602 .

[0126] When the processor 602 executes the program, the automatic driving algorithm verification method based on the cloud-controlled multi-vehicle experimental platform provided in the above embodiment is implemented.

[0127] Furthermore, the cloud-controlled multi-vehicle experimental platform also includes:

[0128] The communication interface 603 is used for communication between the memory 601 and the processor 602 .

[0129] The memory 601 is used to store computer programs that can be executed on the processor 602 .

[0130] The memory 601 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0131] If the memory 601, the processor 602 and the communication interface 603 are implemented independently, the communication interface 603, the memory 601 and the processor 602 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0132] Optionally, in a specific implementation, if the memory 601, the processor 602 and the communication interface 603 are integrated on a chip, the memory 601, the processor 602 and the communication interface 603 can communicate with each other through an internal interface.

[0133] The processor 602 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0134] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform.

[0135] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed, implements the above-mentioned autonomous driving algorithm verification method based on a cloud-controlled multi-vehicle experimental platform.

[0136] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction 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 do not necessarily refer 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, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0137] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0138] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0139] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0140] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one or a combination of multiple of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0141] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0142] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0143] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for verifying an autonomous driving algorithm based on a cloud-controlled multi-vehicle experimental platform, characterized in that: The method is applied to different autonomous driving verification algorithms of cloud-controlled single-vehicle autonomous driving algorithm and cloud-controlled multi-vehicle cooperative control algorithm, and different road verification scenarios of straight roads, curves, and intersections. The switching of execution verification algorithms and verification scenarios is realized through cloud management software, and the content of the cloud-controlled multi-vehicle experimental platform is dynamically adjusted to adapt to different verification requirements, thereby accelerating the development and iteration of autonomous driving algorithms. The method includes the following steps: In the cloud control algorithm adaptation and integration stage, the requirements of the autonomous driving algorithm to be verified are analyzed in depth, so as to adjust the content of the cloud control multi-vehicle experimental platform based on the requirements until the content meets the requirements, wherein the content includes at least one of a physical scaled model vehicle, the number of virtual target vehicles, and a deployment scenario; At the stage when the initial conditions of the cloud control algorithm are met, the real-time monitoring function of the cloud management software is used to determine whether the relative position and posture state of the target vehicle in the requirement meet the requirements of the autonomous driving algorithm to be verified. If not, the relative position and posture state are adjusted in real time by remote control in the cloud until the relative position and posture state meet the requirements; In the cloud-controlled algorithm dynamic execution and performance test stage, the autonomous driving algorithm to be verified is executed, and the expected target vehicle trajectory obtained by the cloud-controlled algorithm calculation and solution in the cloud-controlled multi-vehicle experimental platform is sent to the physical scaled model vehicle and the virtual target vehicle to obtain the verification result of the autonomous driving algorithm to be verified; During the cloud-controlled algorithm comparison test and performance evaluation stage, the cloud-controlled multi-vehicle experimental platform cloud is used to switch the autonomous driving algorithm to be verified to the comparison algorithm to be tested of the autonomous driving algorithm to be verified, and the above-mentioned verification process is repeated. Based on the data collected in real time on the cloud, a comparison result between the autonomous driving algorithm to be verified and the comparison algorithm to be tested is obtained.

2. An autonomous driving algorithm verification system based on a cloud-controlled multi-vehicle experimental platform, characterized in that: include: The cloud-controlled multi-vehicle experimental platform management software is used to select the autonomous driving algorithm to be verified, and select the number of physical scaled model vehicles and virtual target vehicles according to the requirements of different algorithm verifications, and monitor the status information of the target vehicle in real time through a visual interface to obtain the control result of the autonomous driving algorithm to be verified; An autonomous driving algorithm operation platform, used to operate the autonomous driving algorithm to be verified, and generate a reference trajectory corresponding to the target vehicle according to the state information of the target vehicle; The autonomous driving algorithm executes the target vehicle to track the reference trajectory sent from the cloud and respond to cloud control instructions in real time.

Images