Autonomous driving test system and autonomous driving test method

Abstract

The application relates to an automatic driving test system and an automatic driving test method, wherein a virtual simulation end in the automatic driving test system is used for acquiring test requirement parameters and a digital twin scene of an actual vehicle test site, generating vehicle control instructions for a vehicle end to be tested and scene control instructions for a simulation interaction end based on the test requirement parameters and the digital twin scene, the vehicle end to be tested corresponding to a virtual vehicle end, and the simulation interaction end corresponding to a virtual interaction end; the simulation interaction end is used for executing a scene test action in response to the scene control instructions; and the vehicle end to be tested is used for adjusting a vehicle state in response to the vehicle control instructions, so that the ego vehicle adjusts a driving pose according to the scene test action. The embodiment of the application can realize a test scene required for reproducing test requirement parameters of a safe and recoverable target object in a simulation game scene, and replaces dangerous manual operation.

Description

Technical Field

[0001] This application relates to the field of intelligent driving testing technology, and in particular to an autonomous driving testing system and autonomous driving testing method. Background Technology

[0002] With the rapid development of intelligent driving technology, intelligent driving functions are constantly expanding into high-speed and complex scenario domains. Real vehicle testing, as a key link in the verification of intelligent driving technology, is becoming increasingly important.

[0003] However, in real-world testing, traditional intelligent driving algorithms rely on manual operation of the main vehicle and simulated interaction. In extreme and high-risk scenarios (such as a neighboring vehicle cutting in at high speed), the danger increases, which can easily lead to safety accidents. The difficulty of multi-target collaborative control also increases. Manual operation makes it difficult to ensure the consistency of each test scenario, which reduces the credibility and comparability of test results. Furthermore, the pre-test scenario preparation and equipment debugging are time-consuming, while the effective testing time accounts for a small percentage. The overall testing cycle is long and cannot meet the needs of rapid iteration of intelligent driving technology. Summary of the Invention

[0004] To address, or at least partially address, the aforementioned technical problems, this application provides an autonomous driving testing system and an autonomous driving testing method.

[0005] In a first aspect, this application provides an autonomous driving test system, including: a virtual simulation terminal, a test vehicle terminal in a real vehicle test site, and multiple simulated interactive terminals; The virtual simulation terminal is used to acquire test requirement parameters and a digital twin scene of the actual vehicle test site. Based on the test requirement parameters and the digital twin scene, a simulation game scenario is generated. The simulation game scenario includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulation game scenario, vehicle control commands for controlling the vehicle under test terminal in the actual vehicle test site and scene control commands for controlling the simulated interaction terminal in the actual vehicle test site that interacts with the vehicle under test terminal are generated. The vehicle under test terminal corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal. The simulated interactive terminal is used to execute scene test actions in response to the scene control commands; The tested vehicle terminal is used to respond to the vehicle control command and adjust the vehicle state so that the vehicle can adjust its driving posture according to the scenario test action.

[0006] Optionally, the virtual simulation terminal is also used to collect static environmental information of the actual vehicle test site, generate a simulation map file based on the static environmental information, and generate the digital twin scene based on the simulation map file.

[0007] Optionally, the virtual simulation terminal is also used to calibrate the positions of map elements in the simulation map file with the corresponding test elements in the actual vehicle test site.

[0008] Optionally, the cloud is used to send the vehicle control command to the vehicle under test in the real vehicle test site, and to send the scene control command to the simulated interactive terminal in the real vehicle test site. The test vehicle terminal is also used to collect the first vehicle status information of the test vehicle; The simulated interactive terminal is also used to collect the second vehicle status information of the simulated vehicle; The cloud platform is also used to forward the first vehicle status information and the second vehicle status information to the virtual simulation terminal; The virtual simulation terminal is also used to update the simulated game scenario based on the first vehicle status information and the second vehicle status information.

[0009] Optionally, the cloud platform is further configured to determine a safety evaluation score, a reasonableness evaluation score, and a compliance evaluation score based on the first vehicle status information and the second vehicle status information, and output a first test result based on the safety evaluation score, the reasonableness evaluation score, and the compliance evaluation score.

[0010] Optionally, the virtual simulation terminal is further configured to control the virtual vehicle terminal and the virtual interaction terminal to perform virtual simulation in the simulated game scenario, obtain the virtual vehicle status information of the virtual vehicle terminal and the virtual interaction status information of the virtual interaction terminal; and determine the second test result based on the virtual vehicle status information and the virtual interaction status information. The cloud platform is also used to determine simulation test results based on the first test result and the second test result.

[0011] Secondly, this application provides an autonomous driving testing method applied to a virtual simulation platform, the method comprising: Obtain the digital twin scenario of the test requirement parameters and the actual vehicle test site; Based on the test requirement parameters and the digital twin scenario, a simulated game scenario is generated, which includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulated game scenario, vehicle control commands are generated to control the vehicle under test in the real vehicle test site, and scenario control commands are generated to control the simulated interaction terminal that interacts with the vehicle under test in the real vehicle test site. The vehicle under test corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal.

[0012] Optionally, obtain a digital twin scenario of the actual vehicle testing site, including: Collect static environmental information of the actual vehicle test site; A simulation map file is generated based on the static environment information; The digital twin scene is generated based on the simulation map file.

[0013] Optionally, obtaining a digital twin scenario of the actual vehicle testing site also includes: The positions of map elements in the simulation map file are calibrated to the positions of corresponding test elements in the actual vehicle test site.

[0014] Optionally, the method further includes: In the simulated game scenario, the virtual vehicle terminal and the virtual interaction terminal are controlled to perform virtual simulation; Obtain the virtual vehicle status information of the virtual vehicle terminal and the virtual interaction status information of the virtual interaction terminal; The second test result is determined based on the virtual vehicle status information and the virtual interaction status information.

[0015] Thirdly, this application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes a program stored in memory, it implements the autonomous driving test method described in any of the second aspects.

[0016] Fourthly, this application provides a computer-readable storage medium storing a program for an autonomous driving test method, wherein when the program for the autonomous driving test method is executed by a processor, it implements the steps of the autonomous driving test method described in any of the second aspects.

[0017] The beneficial effects of this invention are: This application embodiment obtains test requirement parameters and a digital twin scene of the actual vehicle test site through a virtual simulation terminal. Based on the test requirement parameters and the digital twin scene, a simulated game scenario is generated. The simulated game scenario includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulated game scenario, vehicle control commands for controlling the vehicle under test in the actual vehicle test site and scene control commands for controlling the simulated interaction terminal that interacts with the vehicle under test in the actual vehicle test site are generated. The vehicle under test corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal. The simulated interaction terminal is used to execute scene test actions in response to the scene control commands. The vehicle under test is used to adjust its vehicle state in response to the vehicle control commands, so that the vehicle adjusts its driving posture according to the scene test actions. This invention enables the mapping of control commands generated from a virtual simulation game scenario to a real-world testing ground, with real vehicles responding accordingly. This achieves precise mapping and control of the virtual scenario onto the real-world testing ground. The simulated game scenario controls safe and recoverable targets within the real-world testing ground to reproduce the required test parameters, replacing dangerous manual operations. This solves the problems of difficulty in constructing high-risk scenarios and the high risk of manual operation in real-world testing. Furthermore, the automated testing across the entire chain improves testing efficiency. Additionally, the participation of real vehicles, real roads, and real simulated interactive terminals in the real-world testing ground allows for testing of the vehicle's real sensor recognition performance and vehicle control performance. Finally, this embodiment retains the complete intelligent driving system testing chain from perception to decision-making and planning to vehicle execution, achieving highly complex, efficient, and low-risk intelligent driving real-world testing. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A structural diagram of an autonomous driving test system provided in an embodiment of this application; Figure 2 A communication link diagram of an autonomous driving test system provided in this application embodiment; Figure 3 This is a schematic diagram of a multi-dimensional evaluation model provided in an embodiment of this application; Figure 4A flowchart illustrating an autonomous driving testing method provided in this application embodiment; Figure 5 This is a flowchart of step S102; Figure 6 This is another flowchart for step S102; Figure 7 This is a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] Traditional intelligent driving algorithms, relying on manual operation of the main vehicle and simulated interaction during real-world testing, become increasingly dangerous in extreme and high-risk scenarios (such as adjacent vehicles cutting in at high speeds), easily leading to safety accidents. This increases the difficulty of multi-target collaborative control, and manual operation struggles to ensure consistency across test scenarios, reducing the reliability and comparability of test results. Furthermore, pre-test scenario preparation and equipment debugging are time-consuming, resulting in a low percentage of effective testing time and a long overall testing cycle, failing to meet the demands of rapid iteration in intelligent driving technology. Therefore, this application provides an autonomous driving testing system and method. The aim is to offer a virtual-real mapping test system for real-world vehicle road testing, integrating multi-agent collaborative control, real-time virtual-real mapping, and flexible multi-mode switching to achieve efficient, safe, and high-precision testing and verification of intelligent driving functions.

[0023] This application provides an autonomous driving testing system, such as... Figure 1 As shown, it includes: a virtual simulation terminal and the vehicle under test terminal in the actual vehicle testing site, as well as multiple simulated interactive terminals; The virtual simulation terminal is used to acquire test requirement parameters and a digital twin scene of the actual vehicle test site. Based on the test requirement parameters and the digital twin scene, a simulation game scenario is generated. The simulation game scenario includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulation game scenario, vehicle control commands for controlling the vehicle under test terminal in the actual vehicle test site and scene control commands for controlling the simulated interaction terminal in the actual vehicle test site that interacts with the vehicle under test terminal are generated. The vehicle under test terminal corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal. In this embodiment, the virtual simulation terminal includes virtual simulation software (supporting SimPro, Carsim, and PrescanVTD), a scene generation engine, a scene triggering rule base, a virtual vehicle model, and a data receiving interface; it receives a real vehicle site database and automatically generates virtual static and dynamic scenes; it receives vehicle pose data forwarded from the cloud and updates the virtual vehicle state; it determines scene triggering conditions, generates the next dynamic scene instruction, and sends it to the cloud platform.

[0024] Test requirement parameters refer to the data list of the parameter space used to describe the logical scenario and multiple scenario parameters, including parameters such as the vehicle's speed, longitudinal distance to the adjacent vehicle, speed of the adjacent vehicle, speed of the adjacent vehicle cutting in, cutting time of the adjacent vehicle, lane change direction, initial position coordinates of the vehicle and initial position coordinates of the adjacent vehicle. A digital twin scenario refers to a virtual simulation environment corresponding to a real-world vehicle testing site. The real-world testing site includes roadside sensors, the vehicle under test (VUT), interfering vehicles, and a motion tablet. The VUT is equipped with a terminal, while the interfering vehicles and motion tablets have simulated interactive terminals. The roadside sensors monitor the movement and actual position of targets throughout the site in real time to update the digital twin scenario. The interfering vehicles are stationary real vehicles or rigid dummy vehicles, or real vehicles controlled by driving robots or dummy vehicles towed by a tablet. The motion tablet's platform can carry soft dummy vehicles / humans / motor vehicle models, which move stably along a preset path (straight line / curve / acceleration / deceleration). The digital twin scenario includes a static map, a virtual VUT terminal, and a virtual simulated interactive terminal. The static map is constructed based on the real-world testing site. The virtual VUT terminal corresponds to the VUT in the real-world testing site, and the virtual simulated interactive terminal corresponds to the interfering vehicles or motion tablets in the real-world testing site. Vehicle control commands are used to activate the autonomous driving function of the VUT, start the test, and adjust the vehicle's driving status. Scene control commands are used to control the simulated interactive terminal to perform specific test actions, such as controlling the entry time, movement trajectory and behavior of the simulated interactive terminal, so as to realize the test scenario corresponding to the test requirement parameters.

[0025] For example, a list of requirement parameters that includes multiple test requirement parameters can be shown in the table below: Table 1 List of Requirement Parameters

[0026] The virtual simulation terminal is also used to collect static environmental information of the actual vehicle test site, generate a simulation map file based on the static environmental information, and generate the digital twin scene based on the simulation map file.

[0027] A digital twin scenario refers to a simulation map file generated based on static environmental information collected from a real-world test site. This simulation map file is then used to construct a virtual simulation environment that mirrors the geometry and dynamic logic of the actual test site. The static environmental information refers to the fixed environmental elements within the real-world test site, including the location, size, and type of road boundaries, lane lines, traffic signs, traffic lights, and fixed obstacles. The virtual simulation terminal collects static environmental information from the real-world test site using high-precision measurement equipment, including road boundary width and height, lane line type and spacing, traffic sign location and size, and fixed obstacle location and size.

[0028] The virtual simulation terminal first acquires static environmental information of the actual vehicle test site, then generates a simulation map file based on the static environmental information, and finally generates a digital twin scene based on the simulation map file. For example, the virtual simulation terminal generates a simulation map file based on static information such as road boundaries, lane lines, and traffic signs collected by a high-precision mobile measurement system. Then, based on the simulation map file, it automatically generates a digital twin scene, and then generates a dynamic interactive scene including a virtual test vehicle and a virtual simulation interaction terminal according to the test requirement parameters, thus generating a simulated game scenario.

[0029] In practical applications, high-precision mobile measurement systems, including but not limited to BeiDou + GPS dual-mode positioning modules, LiDAR, high-definition cameras, and road testing equipment, can be used to collect static environmental information, including road boundaries (width, height), lane lines (type, width, spacing), traffic signs (location, size, and sign content), traffic lights (location, number of lights, and control timing interface), and fixed obstacles (such as guardrails, curbs, and trees, location and size). This information is then uploaded to a data preprocessing server via Ethernet for point cloud denoising (removing invalid point clouds caused by rain, fog, and occlusion), image stitching (generating panoramic site images), and data format conversion (converting LiDAR point cloud data to PLY format, image data to JPEG format, and positioning data to the WGS84 coordinate system), forming a standardized real-vehicle site database. A road information collection list containing multiple collected static environmental information items can be exemplified as shown in Table 2 below. Table 2 Road Information Collection List

[0030] Finally, the reference line point set extracted from the point cloud is projected onto the XOY plane. Polynomial coefficients are calculated using cubic spline fitting or the least squares method. A model, including but not limited to a six-parameter model, is then used to convert the registered inertial coordinate system of the point cloud into the OpenDrive reference line coordinate system, ultimately generating an OpenDrive standard virtual simulation map file. The calculation formula for the six-parameter model is as follows: Let the coordinates of the point in the target coordinate system be (X, Y, Z), and the coordinates of the point in the original coordinate system be (x, y, z). The transformation formula is as follows:

[0031] Meaning of each parameter: (ΔX, ΔY, ΔZ): three translation parameters, representing the coordinates of the origin of the original coordinate system in the target coordinate system; k: scaling parameter, used to unify the scale of the two coordinate systems; 3×3 matrix is ​​the rotation matrix.

[0032] The virtual simulation terminal is also used to calibrate the positions of map elements in the simulation map file with the corresponding test elements in the actual vehicle test site.

[0033] The scene generation engine unit performs random sampling based on a natural driving data distribution model. According to the test requirement parameter list, it generates scene parameter combinations and outputs the behavioral trajectories and triggering times of other simulated interactive terminals. Common distribution models for natural driving data include normal distribution and Weiber distribution. The formula for calculating the normal distribution is as follows:

[0034] Where μ is the mean, σ is the standard deviation, and x is the sample value of the driving data.

[0035] The virtual simulation terminal generates two types of control commands based on the simulated game scenario: one is used to activate the vehicle control command of the vehicle under test or adjust the vehicle's driving state, and the other is used to control the simulated interactive terminal to execute specific test actions.

[0036] Scene control instructions include: control type (acceleration / deceleration, steering), target value (target speed, steering angle), and execution time, etc.

[0037] Specifically, the scene generation engine unit generates specific scene parameter combinations based on the logical scene and scene parameter space described in the test requirement parameter list, and on a natural driving data distribution model (including but not limited to normal distribution, Weiber distribution, etc.). It then outputs the behavioral trajectories and triggering times of other simulated interactive terminals and runs them on the virtual simulation terminal (including but not limited to SimPro, VTD, PreScan, etc.). The formula for calculating the normal distribution is as follows:

[0038] Where μ is the mean, σ is the standard deviation, and x is the sample value of the driving data.

[0039] The simulated interactive terminal is used to execute scene test actions in response to the scene control commands; In this embodiment of the application, the simulated interactive terminal is set on the motion plate of the interference vehicle set (a movable plate under the target object in the actual vehicle test site) in the actual vehicle test site, which interacts with the test vehicle. The scenario test action refers to the trajectory running or state switching behavior executed by the simulated interactive terminal according to the control command.

[0040] After receiving the scene control command from the virtual simulation terminal, the simulated interactive terminal responds to the command and executes the corresponding scene test action. There is no need to use a real person to drive a interference vehicle to create dangerous scenarios (such as a neighboring vehicle cutting in at high speed) in the actual vehicle test site. This avoids manual operation, ensures the safety of the test process, the consistency of scene reproduction, and improves test efficiency.

[0041] The vehicle under test is used to respond to the vehicle control command and adjust the vehicle state so that the vehicle can adjust its driving posture according to the scenario test action. In this embodiment, the vehicle under test refers to a real vehicle equipped with the intelligent driving system under test in a real vehicle testing site. Adjusting the vehicle state refers to activating the vehicle's autonomous driving function to put it into the test state, or adjusting the vehicle's driving state, etc.

[0042] After receiving the vehicle control command, the vehicle under test responds to the command and adjusts the vehicle state, enabling the vehicle to automatically adjust its driving speed, steering angle and other driving posture parameters according to the actions executed by the simulated interactive terminal, without human intervention, thus ensuring the safety of the testing process.

[0043] For example, after receiving a vehicle control command, the main test vehicle activates the autonomous driving function. When the target interfering vehicle performs a cutting-in action, the main test vehicle's autonomous driving system automatically makes a decision and performs deceleration or avoidance operations to adjust its driving posture.

[0044] This application embodiment obtains test requirement parameters and a digital twin scene of the actual vehicle test site through a virtual simulation terminal. Based on the test requirement parameters and the digital twin scene, a simulated game scenario is generated. The simulated game scenario includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulated game scenario, vehicle control commands for controlling the vehicle under test in the actual vehicle test site and scene control commands for controlling the simulated interaction terminal that interacts with the vehicle under test in the actual vehicle test site are generated. The vehicle under test corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal. The simulated interaction terminal is used to execute scene test actions in response to the scene control commands. The vehicle under test is used to adjust its vehicle state in response to the vehicle control commands, so that the vehicle adjusts its driving posture according to the scene test actions. This invention enables the mapping of control commands generated from a virtual simulation game scenario to a real-world testing ground, with real vehicles responding accordingly. This achieves precise mapping and control of the virtual scenario onto the real-world testing ground. The simulated game scenario controls safe and recoverable targets within the real-world testing ground to reproduce the required test parameters, replacing dangerous manual operations. This solves the problems of difficulty in constructing high-risk scenarios and the high risk of manual operation in real-world testing. Furthermore, the automated testing across the entire chain improves testing efficiency. Additionally, the participation of real vehicles, real roads, and real simulated interactive terminals in the real-world testing ground allows for testing of the vehicle's real sensor recognition performance and vehicle control performance. Finally, this embodiment retains the complete intelligent driving system testing chain from perception to decision-making and planning to vehicle execution, achieving highly complex, efficient, and low-risk intelligent driving real-world testing.

[0045] In yet another embodiment of this application, as Figure 2 As shown, the autonomous driving testing system also includes: the cloud; The cloud is used to send the vehicle control command to the vehicle under test in the real vehicle test site, and to send the scene control command to the simulated interactive terminal in the real vehicle test site. In this embodiment, the cloud refers to the cloud-based control platform, including an instruction scheduling center, a data processing center, a data storage center, a link status monitoring module, a cloud database, a user interface, and an evaluation system. The hardware configuration utilizes cloud computing servers (e.g., Intel Xeon Gold 6348 CPU, 128GB RAM, 10TB SSD), deployed in a public or private cloud environment. These servers are used to implement instruction scheduling, data forwarding, and link management functions. They primarily receive scenario instructions from the virtual simulation terminal, perform verification and link selection, and then distribute them to the actual vehicle testing site. They also receive data from the actual vehicle, roadside data, and target vehicle, returning it to the virtual simulation terminal for twin updates. Simultaneously, based on the received data, they calculate and output an evaluation report from three dimensions: safety, rationality, and compliance. In addition to the virtual simulation terminal, the tested vehicle terminal, and the simulated interaction terminal, the autonomous driving testing system also includes the cloud as an intermediate hub.

[0046] In other words, the cloud can receive vehicle control commands and scene control commands from the virtual simulation terminal, and send these commands to the vehicle under test and the simulation interaction terminal in the real vehicle test site, respectively.

[0047] In practical applications, the cloud and the actual vehicle site can communicate via dual links (e.g., the main link is 5G-V2X (communication rate ≥1Gbps, latency ≤20ms), and the backup link is fiber optic + 4G (communication rate ≥100Mbps, latency ≤40ms)).

[0048] The test vehicle terminal is also used to collect the first vehicle status information of the test vehicle; The first vehicle status information refers to the motion state and position attitude data of the vehicle under test during the test process, including vehicle speed, steering angle, acceleration and deceleration, and position coordinates obtained based on the global navigation satellite system. For example, the vehicle speed, steering angle, and latitude and longitude position attitude data can be collected by a dual-mode positioning module of Beidou plus global positioning system.

[0049] The simulated interactive terminal is also used to collect the second vehicle status information of the simulated vehicle; The second vehicle status information refers to the motion or working status information collected by the simulated interactive terminal during the execution of test actions.

[0050] The cloud platform is also used to forward the first vehicle status information and the second vehicle status information to the virtual simulation terminal; The virtual simulation terminal is also used to update the simulated game scenario based on the first vehicle status information and the second vehicle status information.

[0051] In this embodiment of the application, updating the digital twin scene refers to synchronizing the data collected in the real test site to the virtual simulation environment so that the state of the vehicle in the virtual scene is consistent with the state of the real vehicle.

[0052] After receiving the first vehicle status information from the vehicle under test and the second vehicle status information from the simulated interaction terminal, the virtual simulation terminal uses these data to synchronously update the digital twin scene, so that the status of the virtual vehicle and the virtual interaction terminal in the virtual scene matches the real status in the actual site.

[0053] After receiving the pose data of the main test vehicle, the virtual simulation terminal uses outlier detection algorithms (including but not limited to the 3σ criterion) to remove jump values ​​and uses mean filtering algorithms (including but not limited to) to smooth the data and reduce noise interference. Then, the processed data is synchronized to the virtual twin scene to update the position and motion state of the virtual vehicle, realizing real-time state mapping from the real vehicle test site to the virtual scene, ensuring the accuracy and reliability of the state of the simulated game scenario. The cloud platform is also used to detect the link quality of wired and wireless communication links, select the communication link with better link quality, send the vehicle control command to the vehicle under test, and send the scene control command to the simulated interactive terminal.

[0054] Specifically, the link status monitoring module can detect the packet loss rate (threshold ≤1%) and latency (threshold ≤50ms) of two links in real time, and automatically select the link with better status to send commands. The controller of the simulated interaction terminal in the real vehicle test site receives scene control commands from the cloud and sends control signals to the power system (throttle, brake) and steering system of the simulated interaction terminal through the CAN bus; the roadside equipment receives scene control commands from the cloud, such as traffic light control commands, and controls the color switching of traffic lights in the test site through the IO interface, with a color switching response time.

[0055] By employing cloud-based dual-link redundancy and dynamic link quality detection mechanisms, automatic optimization and fault switching of communication links are achieved, ensuring low latency and high reliability in the issuance of test commands and avoiding test failures or security risks caused by communication problems.

[0056] This application's embodiments achieve efficient communication decoupling between the virtual simulation terminal and the real vehicle testing site by introducing the cloud as a relay hub for commands and data. The cloud can uniformly manage and schedule test commands and test data, providing scalable architectural support for multi-vehicle collaborative testing and remote testing. Furthermore, through the cloud, the trajectories and behaviors of virtually generated dynamic simulated interactive terminals (such as interfering vehicles or pedestrians) can be mapped in real-time onto the real closed testing site. This allows control of real mobile tablets or interfering vehicles to perform actions and interact with the real vehicle under test. The responses of the vehicle under test and the states of the simulated interactive terminals are collected and transmitted back through the cloud, updating the virtual scene in real-time. This enables real-time correction of the consistency between the virtual and real environments, ensuring that the target vehicle and environmental information strictly follow the set trajectories and state parameters during site testing. Control commands are sent to the site testing in real-time, and the site testing situation is reconstructed in real-time, facilitating observation of the test situation on the platform.

[0057] In another embodiment of this application, the cloud is further configured to determine a safety evaluation score, a reasonableness evaluation score, and a compliance evaluation score based on the first vehicle status information and the second vehicle status information, and output a first test result based on the safety evaluation score, the reasonableness evaluation score, and the compliance evaluation score.

[0058] In this embodiment, the safety evaluation score refers to a quantitative score calculated based on safety indicators such as collision probability, failure recovery time, and collision time. The rationality evaluation score refers to a quantitative score obtained by comparing behavioral indicators such as vehicle speed, steering angle, and acceleration / deceleration with human driving behavior. The compliance evaluation score refers to a quantitative score calculated based on regulatory compliance indicators such as traffic signal recognition accuracy, speed limit compliance rate, and number of violations. The first test result refers to the automated test evaluation report generated according to the report template.

[0059] In practical applications, such as Figure 3 As shown, a multi-dimensional evaluation model encompassing safety, rationality, and compliance can be established. The safety evaluation model includes calculation indicators such as collision probability, failure recovery time, and TTC; the rationality evaluation model includes calculation indicators such as vehicle speed, steering angle, and acceleration / deceleration; and the compliance evaluation model includes calculation indicators such as traffic signal recognition accuracy, speed limit compliance rate, and number of violations. Finally, an automated report output model is established using regular expressions to generate the first test results according to the report template. The evaluation model formula is shown below:

[0060] in, The weight is the weight of the i-th safety indicator (e.g., collision risk weight 0.4, failure response weight 0.3). This represents the actual risk value of the i-th indicator (such as collision probability or failure recovery time).

[0061] The maximum acceptable risk value for the i-th indicator. Let j be the j-th behavioral indicator value of the autonomous driving system (such as vehicle speed or steering angle). Let j be the mean of the j-th indicator of human driving. Let be the standard deviation of the j-th indicator of human driving, k be the number of items that meet regulatory requirements, and K be the total number of all compliance checks.

[0062] The cloud-based automated evaluation module calculates safety, rationality, and compliance scores based on the received first and second vehicle status information. It then uses regular expressions to build an automated report output model and integrates the safety, rationality, and compliance scores according to a preset report template to generate and output the first test result.

[0063] This application embodiment uses a cloud-based automated evaluation module to quantitatively evaluate the testing process from three dimensions: security, rationality, and compliance. This achieves automated generation and standardized output of the first test result, avoiding the subjectivity and inconsistency of manual evaluation and improving the efficiency and objectivity of test evaluation.

[0064] In another embodiment of this application, the virtual simulation terminal is further configured to control the virtual vehicle terminal and the virtual interaction terminal to perform virtual simulation in the simulated game scenario, obtain the virtual vehicle status information of the virtual vehicle terminal and the virtual interaction status information of the virtual interaction terminal; and determine a second test result based on the virtual vehicle status information and the virtual interaction status information. The virtual simulation terminal executes virtual simulation, which can completely synchronize the testing on the real vehicle test site with the simulated game scenario, so that the test process on the test site and the test process in the simulation are consistent, and the test scenario corresponding to the test requirements parameters can be reproduced simultaneously in the virtual game scenario and the real vehicle test site.

[0065] The cloud platform is also used to determine simulation test results based on the first test result and the second test result.

[0066] Based on the aforementioned embodiments, the second test result obtained in the simulated game scenario can be combined with the first test result to generate a second test result. For example, the simulation test result = 0.5 of the first test result + 0.5 of the second test result, or the first test result and the second test result can be compared and a comprehensive simulation test result can be output based on their differences. This enables mutual verification between the real vehicle test site and the simulated game scenario, and allows the site to cover more complex scenarios.

[0067] In summary, the aforementioned embodiments can establish a scenario simulation layer, generate twin virtual map files based on real vehicle site map data collection, automatically generate virtual dynamic scenario files based on scenario parameter lists, and perform scenario simulation. Then, a cloud computing layer is established, which connects to the scenario simulation layer through a cloud application platform to obtain and automatically orchestrate the real-time scenario data of the simulation. Then, through a cloud control platform, the scenario data is parsed to generate trajectory data mapped to site control, and dynamic commands and interactive information are issued to control the interactive behavior and triggering behavior of the local vehicle and the target vehicle on the site, completing the interactive game test. At the same time, the cloud control platform receives the dynamic data of the local vehicle and transmits it to the scenario simulation layer through edge cloud computing units and real-time networks to update the main vehicle information in the virtual simulation scenario, complete the test closed loop, and realize real-time mapping between virtual and real.

[0068] This application can solve the problems of difficulty in constructing complex and risky scenarios for field testing and high risk level of manual testing. At the same time, it realizes the unification of scenario benchmarks, test execution benchmarks and evaluation for simulation testing and field testing, and realizes intelligent driving real vehicle field testing with high complexity, high efficiency and low risk.

[0069] This application consists of: Scene simulation layer → Simulation test (scene library) → Cloud computing layer → Real-time parsing of simulation scenes and distribution to the site → Site test props replicate the simulation scene to execute tests → Feedback of site data to update the state of the scene simulation layer.

[0070] This application comprehensively tests the entire intelligent driving system chain, from environmental perception, sensor fusion, decision-making and planning to vehicle dynamics control and execution. It addresses the challenges of "safe, efficient, and consistent" testing in high-risk scenarios, specifically for real-vehicle field testing. It also resolves the issues of independent field testing and simulation testing, lacking an effective unified testing scenario and benchmark.

[0071] The core technology lies in mapping a massive library of test scenarios and the execution process of scenario testing from the simulated world to a real closed field in real time through a cloud platform, driving the real vehicles being tested and the surrounding traffic participants to completely replicate the complex interaction process in the simulated scenario on a one-to-one basis.

[0072] The pain point it addresses is the high safety risk, poor consistency in scenario reproduction, and low testing efficiency caused by using real people to drive target vehicles to create dangerous scenarios (such as adjacent vehicles cutting in at high speeds) in real-world testing. At the same time, it can better unify the test scenario benchmarks and test execution benchmarks of simulation testing and field testing, which is conducive to improving the confidence and reliability of the three-pillar collaborative test evaluation.

[0073] In another embodiment of this application, an autonomous driving testing method is also provided, applied to a virtual simulation terminal, such as... Figure 4 As shown, the method includes: Step S101: Obtain the test requirement parameters and the digital twin scene of the actual vehicle test site; Step S102: Generate a simulated game scenario based on the test requirement parameters and the digital twin scenario. The simulated game scenario includes a virtual vehicle terminal and a virtual interaction terminal. Step S103: Based on the simulated game scenario, generate vehicle control commands for controlling the vehicle under test in the real vehicle test site and scene control commands for controlling the simulated interaction terminal that interacts with the vehicle under test in the real vehicle test site. The vehicle under test corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal. This application embodiment generates a simulated game scenario and maps the control commands generated by the virtual simulated game scenario to the real site, which is then responded to by real vehicles. This achieves precise mapping and control of the virtual scenario to the real vehicle test site, so that the simulated game scenario can be used to control the test scenario required to reproduce the test requirements parameters of safe and recoverable targets in the real vehicle test site. This replaces dangerous manual operation and solves the problems of difficulty in constructing high-risk scenarios and high risks of manual operation in real vehicle testing. Moreover, it improves test efficiency through full-link automatic testing.

[0074] In another embodiment of this application, a digital twin scenario corresponding to the actual vehicle test site is generated based on the test requirement parameters, such as... Figure 5 As shown, it includes: Step S201: Collect static environmental information of the actual vehicle test site; In this embodiment of the application, static environmental information refers to fixed environmental elements in the actual vehicle test site, including the location, size and type information of road boundaries, lane lines, traffic signs, traffic lights and fixed obstacles.

[0075] In this step, the virtual simulation terminal collects static environmental information of the actual vehicle test site through high-precision measurement equipment, including road boundary width and height, lane line type and width spacing, traffic sign position and size, fixed obstacle position and size, etc.

[0076] For example, the virtual simulation terminal uses a dual-mode positioning module of Beidou plus global positioning system, lidar and high-definition camera to collect static environmental information such as the width and height of road boundaries, lane line type and width spacing, traffic sign location and size, and fixed obstacle location and size.

[0077] Step S202: Generate a simulation map file based on the static environment information; In this embodiment of the application, the simulated map file refers to a map file that conforms to the Open Dynamic Traffic Environment Exchange Series Standard.

[0078] In this step, the virtual simulation terminal uploads the collected static environment information to the data preprocessing server via Ethernet for point cloud denoising, image stitching, and data format conversion. The six-parameter model is used to convert the inertial coordinate system after point cloud registration into the reference line coordinate system of the Open Dynamic Traffic Environment Exchange series, and finally generates the Open Dynamic Traffic Environment Exchange standard virtual simulation map file.

[0079] For example, the virtual simulation terminal converts lidar point cloud data into a 3D format, image data into the Joint Image Experts Group (JEAG) format, and positioning data into a global geospatial information coordinate system. Through a six-parameter model, coordinate transformation is performed to generate an open dynamic traffic environment exchange standard virtual simulation map file.

[0080] After generating the simulation map file, the positions of the map elements in the simulation map file can be calibrated with the corresponding test elements in the actual vehicle test site to ensure twin consistency.

[0081] Step S203: Generate the digital twin scene based on the simulation map file.

[0082] In this embodiment, the virtual vehicle under test refers to a virtual object representing the real vehicle under test in the digital twin scenario. The virtual simulated interaction terminal refers to a virtual object representing the real simulated interaction terminal in the digital twin scenario.

[0083] In this step, the virtual simulation terminal, based on the logical scenario and parameter space in the test requirements parameters, adds a virtual vehicle under test and a virtual simulation interaction terminal to the simulation map file to generate a complete digital twin scenario.

[0084] For example, the virtual simulation terminal adds virtual master vehicles and virtual target vehicles to the virtual simulation map file of the Open Dynamic Traffic Environment Exchange Standard based on the test requirements parameters of the neighbor vehicle entry scenario, sets the initial position, entry speed and triggering time of the virtual target vehicle, and generates a digital twin scenario.

[0085] This application embodiment achieves a refined mapping from the real site to the virtual scene through three sub-steps: collecting static environmental information, generating simulation map files, and generating a digital twin scene containing virtual vehicles. This ensures the geometric accuracy and dynamic logic authenticity of the twin scene.

[0086] In another embodiment of this application, a digital twin scenario corresponding to the actual vehicle test site is generated based on the test requirement parameters, such as... Figure 6 As shown, it includes: Step S301: In the simulated game scenario, control the virtual vehicle terminal and the virtual interaction terminal to perform virtual simulation; Step S302: Obtain the virtual vehicle status information of the virtual vehicle terminal and the virtual interaction status information of the virtual interaction terminal; Step S303: Determine the second test result based on the virtual vehicle status information and the virtual interaction status information.

[0087] In this embodiment of the application, virtual vehicle terminals and virtual interactive terminals are controlled to perform virtual simulation in a simulated game scenario. This allows the testing at the real vehicle test site to be completely synchronized with the simulated game scenario, ensuring that the test process at the test site is consistent with the simulated test process. Both the virtual game scenario and the real vehicle test site can reproduce the test scenario corresponding to the test requirements parameters, so as to compare the test results.

[0088] In another embodiment of this application, an electronic device is also provided, characterized in that it includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes the program stored in the memory, it implements the autonomous driving test method described in any of the foregoing method embodiments.

[0089] The electronic device provided in this invention enables the processor to execute programs stored in the memory, realizing a complete testing process from requirements to instructions to closed-loop feedback. It can complete automated testing of complex scenarios without manual intervention, thereby improving testing efficiency and security.

[0090] The communication bus 1140 mentioned in the above-mentioned electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus 1140 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0091] The communication interface 1120 is used for communication between the above-mentioned electronic device and other devices.

[0092] The memory 1130 may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0093] The processor 1110 mentioned above can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0094] In another embodiment of this application, a computer-readable storage medium is also provided, on which a program for an autonomous driving test method is stored, wherein when the program for the autonomous driving test method is executed by a processor, it implements the steps of the autonomous driving test method described in any of the foregoing method embodiments.

[0095] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0096] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An autonomous driving testing system, characterized in that, include: The virtual simulation terminal and the vehicle under test terminal and multiple simulated interactive terminals in the real vehicle test site; The virtual simulation terminal is used to acquire test requirement parameters and a digital twin scene of the actual vehicle test site. Based on the test requirement parameters and the digital twin scene, a simulation game scenario is generated. The simulation game scenario includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulation game scenario, vehicle control commands for controlling the vehicle under test terminal in the actual vehicle test site and scene control commands for controlling the simulated interaction terminal in the actual vehicle test site that interacts with the vehicle under test terminal are generated. The vehicle under test terminal corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal. The simulated interactive terminal is used to execute scene test actions in response to the scene control commands; The tested vehicle terminal is used to respond to the vehicle control command and adjust the vehicle state so that the vehicle can adjust its driving posture according to the scenario test action.

2. The autonomous driving test system according to claim 1, characterized in that, The virtual simulation terminal is also used to collect static environmental information of the actual vehicle test site, generate a simulation map file based on the static environmental information, and generate the digital twin scene based on the simulation map file.

3. The autonomous driving test system according to claim 2, characterized in that, The virtual simulation terminal is also used to calibrate the positions of map elements in the simulation map file with the corresponding test elements in the actual vehicle test site.

4. The autonomous driving test system according to claim 1, characterized in that, Also includes: Cloud; The cloud is used to send the vehicle control command to the vehicle under test in the real vehicle test site, and to send the scene control command to the simulated interactive terminal in the real vehicle test site. The test vehicle terminal is also used to collect the first vehicle status information of the test vehicle; The simulated interactive terminal is also used to collect the second vehicle status information of the simulated vehicle; The cloud platform is also used to forward the first vehicle status information and the second vehicle status information to the virtual simulation terminal; The virtual simulation terminal is also used to update the simulated game scenario based on the first vehicle status information and the second vehicle status information.

5. The autonomous driving test system according to claim 4, characterized in that, The cloud platform is also used to determine a safety evaluation score, a reasonableness evaluation score, and a compliance evaluation score based on the first vehicle status information and the second vehicle status information, and to output a first test result based on the safety evaluation score, the reasonableness evaluation score, and the compliance evaluation score.

6. The autonomous driving test system according to claim 5, characterized in that, The virtual simulation terminal is also used to control the virtual vehicle terminal and the virtual interaction terminal to perform virtual simulation in the simulated game scenario, obtain the virtual vehicle status information of the virtual vehicle terminal and the virtual interaction status information of the virtual interaction terminal; and determine the second test result based on the virtual vehicle status information and the virtual interaction status information. The cloud platform is also used to determine simulation test results based on the first test result and the second test result.

7. An autonomous driving testing method, characterized in that, Applied to a virtual simulation terminal, the method includes: Obtain the digital twin scenario of the test requirement parameters and the actual vehicle test site; Based on the test requirement parameters and the digital twin scenario, a simulated game scenario is generated, which includes a virtual vehicle terminal and a virtual interaction terminal. Based on the simulated game scenario, vehicle control commands are generated to control the vehicle under test in the real vehicle test site, and scenario control commands are generated to control the simulated interaction terminal that interacts with the vehicle under test in the real vehicle test site. The vehicle under test corresponds to the virtual vehicle terminal, and the simulated interaction terminal corresponds to the virtual interaction terminal.

8. The autonomous driving testing method according to claim 7, characterized in that, Obtain a digital twin scenario of the actual vehicle testing site, including: Collect static environmental information of the actual vehicle test site; A simulation map file is generated based on the static environment information; The digital twin scene is generated based on the simulation map file.

9. The autonomous driving testing method according to claim 8, characterized in that, Obtaining a digital twin scenario of a real vehicle testing site also includes: The positions of map elements in the simulation map file are calibrated to the positions of corresponding test elements in the actual vehicle test site.

10. The autonomous driving testing method according to claim 7, characterized in that, The method further includes: In the simulated game scenario, the virtual vehicle terminal and the virtual interaction terminal are controlled to perform virtual simulation; Obtain the virtual vehicle status information of the virtual vehicle terminal and the virtual interaction status information of the virtual interaction terminal; The second test result is determined based on the virtual vehicle status information and the virtual interaction status information.

11. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor, when executing a program stored in memory, implements the autonomous driving test method according to any one of claims 7-10.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program for an autonomous driving test method, which, when executed by a processor, implements the steps of the autonomous driving test method according to any one of claims 7-10.

Images