A virtual testing method based on a driving simulator

Abstract

The application relates to a virtual test method based on a driving simulator, which comprises the following steps: constructing a static environment part of a virtual test space and configuring the test vehicle, introducing the test vehicle and an artificially operated interference vehicle into the virtual test space, and completing the preparation before the test; performing the test in an actual test environment, controlling the interference vehicle to perform confrontation interference on the test vehicle through the driving simulator, obtaining recorded data of the test vehicle; constructing a dynamic part of the virtual test space based on the recorded data, forming a complete driving scene, evaluating the driving scene, and completing the virtual test on the driving control software. Compared with the prior art, the application has the advantages of high fitting degree, virtual-real combination and the like.

Description

Technical Field

[0001] This invention relates to the field of driving test technology, and in particular to a virtual testing method based on a driving simulator. Background Technology

[0002] With the advancement of computer technology, the application of autonomous driving technology in vehicles has increased significantly. However, before autonomous driving technology can be fully implemented, extensive testing is still needed to ensure the safety performance of autonomous vehicles. The 2020 China Autonomous Driving Simulation Blue Book points out that before autonomous vehicles can be truly commercialized, their algorithms need to be tested and optimized. However, current road testing for vehicles still suffers from insufficient road driving data and cannot meet the safety testing requirements under dangerous conditions. Therefore, testing autonomous driving technology through virtual scenario simulation has become an important technical approach.

[0003] Virtual scenes can be categorized into static and dynamic scenes based on their environmental conditions. Static scenes include static elements related to vehicle movement, such as roads (including materials, lane markings, speed bumps, etc.) and static traffic elements (including traffic signs, streetlights, stations, tunnels, surrounding buildings, etc.). Dynamic scenes include dynamic environmental elements such as dynamic signage facilities and communication environment information, as well as traffic participants (including motor vehicle behavior, non-motor vehicle behavior, pedestrian behavior, etc.), weather changes (rain, snow, fog, etc.), and time changes (mainly changes in lighting at different times). By combining dynamic and static scenes, a testing environment identical to the real world can be constructed in virtual space.

[0004] Autonomous driving test scenarios can be categorized into normal scenarios and extreme scenarios based on their frequency of occurrence. Currently, autonomous driving software algorithms are relatively mature in testing normal scenarios, but when facing testing in extreme scenarios, there are often problems such as difficulty in obtaining or reproducing the scenarios.

[0005] Most existing scene construction methods are based on real-world data, which inevitably leads to problems such as the inability to collect data or incomplete data in extreme cases. This results in virtual scenes not being generated accurately and dangerous scenarios not being reproduced. At the same time, virtual scenes generated by generalizing from real-world data also suffer from low confidence and poor fit to reality. Furthermore, scenes generated from virtual data often fail to accurately represent the behavior and state of real vehicles, resulting in a "separation of virtual and real" in the motion state of the vehicle model within the formed virtual scene. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a virtual testing method based on a driving simulator. This method uses a driving simulator to control a disruptive vehicle to counteract interference with the test vehicle, collects data during the testing process, and evaluates the data to complete the testing of the driving control software.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] This invention provides a virtual testing method based on a driving simulator, comprising the following steps:

[0009] Construct the static environment portion of the virtual test space and configure it for the test vehicle. Introduce the test vehicle and manually controlled interference vehicles into the virtual test space to complete the preparations before the test.

[0010] In a real-world testing environment, the driving control software of the test vehicle is tested. A driving simulator is used to control the interfering vehicle to counteract the interference of the test vehicle, and the recorded data of the test vehicle is obtained.

[0011] Based on the recorded data, a dynamic portion of the virtual test space is constructed to form a complete driving scenario, which is then filtered.

[0012] As a preferred technical solution, the construction of the static environment portion of the virtual test space and the configuration for the test vehicle includes the following steps:

[0013] Configure the sand table roads in the actual test space according to the test space requirements, and construct a static scene in the virtual test space based on the sand table roads;

[0014] The driving control software was loaded onto the test vehicle in the actual test environment.

[0015] As a preferred technical solution, static scenes in a virtual test space can be constructed by modeling the environment in Carla or Sumo.

[0016] As a preferred technical solution, the following steps are also included:

[0017] Modify the roads in the sand table and update the static scene accordingly.

[0018] As a preferred technical solution, the anti-interference measures for the test vehicle include the following steps:

[0019] The test vehicle was subjected to counter-interference tactics including malicious cutting off other vehicles, sharp turns, and emergency braking.

[0020] As a preferred technical solution, obtaining the recorded data of the test vehicle includes the following steps:

[0021] The system acquires sensor data integrated on the vehicle, sand table road test camera data, sand table UWB data, and input data from the driving simulator. Based on the sensor data, it acquires road information from the vehicle's perspective. Based on the sand table road test camera data, it acquires vehicle information measured on the road. Based on the sand table UWB data, it acquires the location information of the test vehicle.

[0022] The recorded data is obtained after integration and packaging.

[0023] As a preferred technical solution, the screening for the driving scenario includes the following steps:

[0024] For the newly generated driving scenario, the model is used to score the performance of the test vehicle during the test process, and it is determined whether the score exceeds a preset threshold. If so, the newly generated driving scenario is saved.

[0025] As a preferred technical solution, the driving scenarios are stored in a preset scenario library.

[0026] As a preferred technical solution, the model described is a responsibility-sensitive safety model.

[0027] As a preferred technical solution, the driving simulator controls multiple interfering vehicles to form a traffic flow, thereby counteracting interference to the test vehicle.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] (1) High degree of fit: The present invention first completes the construction of the static environment of the virtual test space before testing, and then conducts the test in the actual test environment. By using a driving simulator to control the interference vehicle to counteract the interference of the test vehicle, the recorded data during the test process is collected to construct the dynamic part of the virtual test space, forming a complete driving scenario. Finally, the generated driving scenario is evaluated to complete the test of the driving control software. Compared with the traditional scenario construction method based on real data, which has the problem that data cannot be collected or is incomplete in extreme cases, resulting in the virtual scenario not being accurately generated and dangerous scenarios not being reproduced, the virtual scenario generated by generalizing from real data also has the problem of low confidence and low degree of fit with reality. The present invention uses a driving simulator to control the interference vehicle to counteract the interference of the test vehicle, which can simulate extreme scenarios and has a high degree of fit with reality.

[0030] (2) "Virtual-real combination": Traditional methods often fail to accurately express the behavior and state of real vehicles through virtual data. In the virtual scene, the motion state of the vehicle model will be "separated from the virtual". The driving scene obtained by this invention includes a scene part based on the actual sand table and a dynamic scene part based on the recorded data. Combining the two parts to obtain a complete driving scene can accurately express the behavior and state of real vehicles. Using the filtered driving scene to optimize the driving control software can improve the performance of the driving control software. Attached Figure Description

[0031] Figure 1 This is a flowchart of the virtual testing method based on a driving simulator in the embodiment. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0033] Example 1

[0034] like Figure 1 This embodiment provides a virtual testing method based on a driving simulator for testing driving control software. In a real-world sandbox environment, a human-controlled vehicle is used to counteract interference with a test vehicle equipped with the autonomous driving test software, obtaining scenarios under specific conditions. This process is then projected into a virtual space, created as a scenario, and stored in a database for future improvements and optimizations to the autonomous driving software. This method includes the following steps:

[0035] Step S1: Load the autonomous driving test software onto the test vehicle in the real sand table. Utilize the configurability of the real sand table to change the road conditions in the real sand table and simultaneously synchronize the road updates to the virtual space to test the operation of the autonomous driving test vehicle under different road conditions.

[0036] Step S2: Configure the sand table road environment according to environmental requirements, and complete the construction of the static scene part in the virtual scene (opendrive) based on the sand table. Model the environment in the virtual space (such as Carla, Sumo) based on the real sand table environment, and ensure the consistency between the virtual scene and the real sand table through virtual-real mapping.

[0037] Step S3: Load the test vehicle and the autonomous vehicle in the virtual space.

[0038] Step S4: The test vehicle undergoes testing. Simultaneously, a driving simulator manipulates the autonomous vehicle to counteract interference (such as malicious lane-changing or emergency braking) to test the functionality and stability of the autonomous driving software. When such interference occurs, information from the test vehicle is recorded and collected using sensors integrated into the vehicle, roadside cameras on the sand table, and UWB on the sand table. This data is then integrated and packaged with the input from the driving simulator. All data packets are sent to the virtual scene to construct a portion of the dynamic scene (openscenario) and generate a complete driving scenario, thus constructing the entire driving condition.

[0039] By mounting cameras on a sand table vehicle and connecting the camera signals to a driving simulator display, users can manually control the autonomous vehicle to counteract interference, such as cutting off other vehicles, making sharp turns, and performing emergency stops. Ultra-wideband (UWB) wireless communication technology is used to acquire real-time vehicle position information within the sand table. Data is collected from cameras positioned on the actual sand table vehicle and from cameras along the sand table roads to simultaneously obtain road information from the vehicle's perspective and vehicle information measured from the road. The input information from the driving simulator and the sensor data from the actual sand table vehicle are combined to obtain a complete vehicle status report.

[0040] Preferably, multiple driving simulators can simultaneously control multiple self-controlled vehicles on a sand table to form traffic flow and simulate driving scenarios that may occur in reality.

[0041] Step S5: The generated scenarios are scored and evaluated, a score threshold is set, and scenarios that meet the requirements are stored in the scenario library. Usable scenarios are extracted from the scenario library to optimize and improve the autonomous driving test software.

[0042] Preferably, a Responsibility-Sensitive Safety (RSS) model is used to screen the constructed scenarios.

[0043] This method uses a driving simulator to control a disruptive vehicle to counteract interference on the test vehicle, enabling simulation of extreme scenarios with a high degree of realism. Furthermore, the driving scenarios acquired in this invention include both scenarios based on an actual sand table and dynamic scenarios based on recorded data. Combining these two parts yields a complete driving scenario that accurately represents the behavior and state of a real vehicle. Optimizing the driving control software using the selected driving scenarios improves its performance.

[0044] Example 2

[0045] This embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, the one or more programs including instructions for executing the virtual testing method based on a driving simulator as described in Embodiment 1.

[0046] Example 3

[0047] This embodiment provides a computer-readable storage medium, characterized in that it includes one or more programs executable by one or more processors of an electronic device, the one or more programs including instructions for performing the virtual testing method based on a driving simulator as described in Embodiment 1.

[0048] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A virtual testing method based on a driving simulator, characterized in that, Includes the following steps: Construct the static environment portion of the virtual test space and configure it for the test vehicle. Introduce the test vehicle and manually controlled interference vehicles into the virtual test space to complete the preparations before the test. In a real-world testing environment, the driving control software of the test vehicle is tested. A driving simulator is used to control the interfering vehicle to counteract the interference of the test vehicle, and the recorded data of the test vehicle is obtained. Based on the recorded data, a dynamic portion of the virtual test space is constructed to form a complete driving scenario, which is then filtered. The static environment component of constructing the virtual test space and configuring it for the test vehicle includes the following steps: Configure the sand table roads in the actual test space according to the test space requirements, and construct a static scene in the virtual test space based on the sand table roads; The driving control software was loaded onto the test vehicle in the actual test environment. Countering interference on the test vehicle includes the following steps: The test vehicle was subjected to counter-interference tactics including malicious cutting off, sharp turns, and emergency braking. Obtaining the recorded data from the test vehicle includes the following steps: The system acquires sensor data integrated on the vehicle, sand table road test camera data, sand table UWB data, and input data from the driving simulator. Based on the sensor data, it acquires road information from the vehicle's perspective. Based on the sand table road test camera data, it acquires vehicle information measured on the road. Based on the sand table UWB data, it acquires the location information of the test vehicle. The recorded data is obtained after integration and packaging. The filtering process for the driving scenario includes the following steps: For the newly generated driving scenario, the model is used to score the performance of the test vehicle during the test process. It is then determined whether the score exceeds a preset threshold. If so, the newly generated driving scenario is saved. The driving scenarios are stored in a preset scenario library.

2. The virtual testing method based on a driving simulator according to claim 1, characterized in that, Static scenes in a virtual test space can be constructed by modeling the environment in Carla or Sumo.

3. The virtual testing method based on a driving simulator according to claim 1, characterized in that, It also includes the following steps: Modify the roads in the sand table and update the static scene accordingly.

4. The virtual testing method based on a driving simulator according to claim 1, characterized in that, The model described is a responsibility-sensitive security model.

5. The virtual testing method based on a driving simulator according to claim 1, characterized in that, The driving simulator controls multiple interfering vehicles to form a traffic flow, thereby counteracting interference with the test vehicle.

Images