Safety test method and device for autonomous vehicle

Abstract

The invention relates to a safety test method and device for an autonomous vehicle. The method comprises the steps that in the embodiment of the application, the multi-level test schemes corresponding to the automatic driving grading standard are determined, and at least part of the multi-level test schemes comprise test content in a dangerous state scene; simulating a scene corresponding to the multi-level test scheme through a driving simulator, and obtaining test information corresponding to multiple levels; the test information comprises interaction information of the driver and the tested vehicle in the test process, and the interaction information at least comprises interaction information in the test process in the dangerous state scene. According to the method, control and switching of different levels of automatic driving are realized, human-vehicle dynamic interaction of the automatic driving automobile in a dangerous state scene is carried out under the condition that the tested safety is ensured, the efficiency is improved through a dangerous state acceleration test, and acquisition of human-vehicle interaction characteristics in the dangerous state scene is realized.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a safety testing method and device for an autonomous driving vehicle. Background Art

[0002] As vehicles become increasingly intelligent, driving control is shifting from humans to machines. In real-world traffic environments, unavoidable collision scenarios present a key challenge. In these extreme situations, the driver's response and the collaborative mechanisms of the intelligent driving system directly impact the probability of a collision and the severity of the damage. Therefore, analyzing driver response in these critical scenarios is crucial to optimizing autonomous driving technology.

[0003] Traditional technologies usually focus on monitoring and identifying drivers' distracted driving, fatigue driving, drunk driving and other conditions.

[0004] However, traditional technologies lack a safety testing method for autonomous driving vehicles that collects human-vehicle interaction characteristics in dangerous scenarios. Summary of the Invention

[0005] Based on this, it is necessary to provide a safety testing method and device for autonomous driving vehicles that can collect human-vehicle interaction characteristics in dangerous scenarios to address the above technical problems.

[0006] In a first aspect, the present application provides a safety testing method for an autonomous driving vehicle, the method comprising:

[0007] Determining a multi-level test plan corresponding to the automated driving classification standard, wherein at least some of the multi-level test plans include: test content for dangerous scenarios;

[0008] The driving simulator simulates the scenarios corresponding to the multi-level test scheme and obtains test information corresponding to the multi-level test scheme; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios,

[0009] In one embodiment, at least one level of the multi-level test scheme includes: test content in a non-dangerous scenario; simulating the multi-level test scheme with a driving simulator to obtain test information corresponding to the multiple levels includes:

[0010] simulating the non-dangerous scenario and the dangerous scenario by the driving simulator;

[0011] In the non-dangerous scenario and the dangerous scenario, the driving state information and vehicle state information of the driver are collected to obtain test information corresponding to multiple levels.

[0012] In one embodiment, the multi-level testing scheme includes: a manual driving test scheme and a multi-level autonomous driving test scheme, and the simulating of the dangerous scenario by a driving simulator includes:

[0013] Simulating the dangerous scenario in the manual driving test scheme by the driving simulator;

[0014] The driving simulator is used to simulate dangerous scenarios in any level of autonomous driving test scheme.

[0015] In one embodiment, the method further comprises:

[0016] A multi-level autonomous driving human-vehicle interaction response dataset is constructed based on the test information and the basic information of the driver.

[0017] In one embodiment, after constructing a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the basic information of the driver, the method further includes:

[0018] Determining a test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset;

[0019] Obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver;

[0020] Based on the baseline collision rate and the test collision rate, determine whether the vehicle under test meets the autonomous driving safety requirements.

[0021] In one embodiment, the baseline collision rate and the test collision rate correspond to a target dangerous situation type; and determining whether the tested vehicle meets the autonomous driving safety requirements based on the baseline collision rate and the test collision rate includes:

[0022] When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, determining that the autonomous driving safety requirement is not met;

[0023] Based on the difference, the automatic driving behavior of the tested vehicle under the target dangerous state type is optimized and controlled.

[0024] In one embodiment, after constructing a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the basic information of the driver, the method further includes:

[0025] Obtaining safety assessment parameters from the response data set, the safety assessment parameters including at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter;

[0026] Based on the safety assessment parameters, a safety score of the tested vehicle is determined.

[0027] In one embodiment, determining a multi-level testing scheme corresponding to the autonomous driving classification standard includes:

[0028] Determining the vehicle functions corresponding to each level of autonomous driving according to the autonomous driving classification standards;

[0029] The multi-level testing scheme is determined according to the vehicle functions corresponding to each level of autonomous driving.

[0030] In a second aspect, the present application also provides a safety testing device for an autonomous driving vehicle, comprising:

[0031] A first determining module is configured to determine a multi-level test scheme corresponding to the autonomous driving classification standard, wherein at least some of the multi-level test schemes include: test content under dangerous scenarios;

[0032] The first acquisition module is used to simulate the scenarios corresponding to the multi-level test scheme through a driving simulator and obtain test information corresponding to the multiple levels; the test information includes: interaction information between the driver and the vehicle under test during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

[0033] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0034] Determining a multi-level test plan corresponding to the automated driving classification standard, wherein at least some of the multi-level test plans include: test content for dangerous scenarios;

[0035] The scenarios corresponding to the multi-level test scheme are simulated by a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

[0036] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0037] Determining a multi-level test plan corresponding to the automated driving classification standard, wherein at least some of the multi-level test plans include: test content for dangerous scenarios;

[0038] The scenarios corresponding to the multi-level test scheme are simulated by a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

[0039] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:

[0040] Determining a multi-level test plan corresponding to the automated driving classification standard, wherein at least some of the multi-level test plans include: test content for dangerous scenarios;

[0041] The scenarios corresponding to the multi-level test scheme are simulated by a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

[0042] The aforementioned safety testing method and apparatus for autonomous vehicles determine a multi-level test scheme corresponding to the autonomous driving classification standard. At least some of the multi-level test schemes include: test content for hazardous scenarios; scenarios corresponding to the multi-level test schemes are simulated via a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test, with the interaction information at least including interaction information during the test in hazardous scenarios. By simulating scenarios corresponding to different levels of test schemes via a driving simulator, control and switching between different levels of autonomous driving can be achieved, ensuring the safety of the test subject while conducting dynamic human-vehicle interaction in hazardous scenarios. Efficiency is improved through hazardous acceleration testing, enabling the collection of human-vehicle interaction characteristics in hazardous scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 A diagram illustrating an application environment of a safety testing method for an autonomous driving vehicle according to an embodiment;

[0045] Figure 2 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to an embodiment;

[0046] Figure 3 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0047] Figure 4 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0048] Figure 5 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0049] Figure 6 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0050] Figure 7 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0051] Figure 8 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0052] Figure 9 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0053] Figure 10 1 is a flow chart of a safety testing method for an autonomous driving vehicle according to another embodiment;

[0054] Figure 11 1 is a structural block diagram of a safety testing device for an autonomous driving vehicle according to an embodiment;

[0055] Figure 12 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0057] Road traffic scenarios are complex and ever-changing. The dynamic interaction of the three elements of "driver, vehicle, and road" together forms a highly dynamic, strongly coupled, and nonlinear generalized dynamic system. Safety-critical scenarios, distinct from normal driving scenarios, are those in which sources of traffic conflict exist and could potentially escalate into collisions. These critical scenarios play a crucial role in safety risks. When a critical scenario occurs, the driver goes through three primary stages: hazard perception, driving decision-making, and collision avoidance execution. These processes control the interaction between the vehicle and the hazard source and influence whether a collision occurs. If a collision is not avoided, the human body experiences impact loads and can sustain injuries during the collision phase.

[0058] As the level of automotive intelligence increases, the primary driver of driving control is gradually shifting from humans to machines. However, the goal of automotive intelligence development remains to assist or replace the driver in achieving optimal vehicle control and safety. In the long term, before intelligent vehicles achieve industrial application and autonomous driving, drivers and intelligent driving systems will share vehicle control and road access. Therefore, understanding human drivers' ability to understand and interact with autonomous vehicles at different levels, and assessing the safety benefits of autonomous vehicles at different levels, will guide the development of autonomous vehicle technology and the formulation of social policies by public managers. It is also crucial to clarify the interaction mechanisms between drivers and intelligent driving systems.

[0059] Existing autonomous driving technologies primarily focus on interactive scenarios within normal driving tasks, such as spatial perception and trajectory planning. However, research has shown that with the continued development of intelligent vehicles and technological advancements, unavoidable collisions will become a key challenge in real-world traffic environments. In these extreme situations, the driver's response and the collaborative mechanisms of the intelligent driving system will directly influence the probability of a collision and the severity of the damage.

[0060] Traditional testing methods have made some progress in monitoring and identifying drivers in specific conditions, such as fatigue, distracted driving, and drunk driving. For example, these methods use data collected from on-board sensors, wearable devices, and mobile terminals to determine a driver's mental state and attention level. However, these studies focus more on the driver's state of mind itself and less on the driver's interaction with different levels of automated driving systems in hazardous scenarios. Under hazardous traffic conditions, drivers' decision-making behavior can differ significantly when operating in automated driving systems of different levels, or even in automated driving systems of the same level but with different functions. Therefore, exploring drivers' decision-making patterns, risk perception, and coordination mechanisms with automated driving systems under hazardous conditions is crucial for improving the safety and reliability of intelligent driving technology.

[0061] While existing driver assistance systems (such as Autonomous Emergency Braking (AEB) and Forward Collision Warning (FCW)) can provide warnings or proactively intervene in emergency situations to mitigate collision risks, their decision-making logic is typically based on fixed triggering conditions rather than dynamically adjusting to the individual driver's perceptual characteristics. To strike a balance between safety and comfort, these systems often reduce trigger sensitivity to minimize false alarms. However, this leads to potential false alarms, whereby the system fails to intervene in certain emergency situations, potentially compromising accident avoidance effectiveness. This reflects a failure to fully consider the driver's individual risk perception and response characteristics in their design. This results in poor alignment between warning and intervention triggering logic and the driver's actual operating habits and cognitive patterns, potentially leading to potential risks and unintended collisions. Therefore, implementing intelligent, tailored interaction based on the driver's perceptual characteristics in dangerous interaction situations, enabling the driver assistance system to dynamically adjust its triggering strategy and form a more efficient collaborative mechanism with the driver, is crucial for improving the safety, reliability, and user acceptance of autonomous driving systems. At the same time, most research in the field of traffic and automobile safety focuses on driver control or traffic flow interaction. There is a lack of repeated experiments with clearly defined scenarios, making it difficult to accurately restore and compare the scenarios, thereby studying the impact of the driver's perceptual decision-making factors on dangerous interactions.

[0062] Furthermore, traditional technologies primarily derive automobile safety data from critical traffic scenarios from traditional data sources such as accident investigation and reconstruction datasets and road aerial photography datasets. Accident statistics datasets primarily focus on post-crash investigation and reconstruction, allowing for analysis of collision severity and occupant injuries. However, since they primarily record the aftermath of an accident and lack the dynamic evolution preceding the collision, they struggle to accurately reconstruct the key decision-making behaviors of the driver and vehicle during critical traffic scenarios. Meanwhile, road aerial photography datasets effectively capture vehicle trajectories and interactions, providing some support for autonomous driving testing. However, these datasets contain limited information dimensions, focusing primarily on motion video, making them difficult to fully capture critical traffic conditions. Existing data lacks granular data at critical moments before and during the collision process, making it difficult for researchers to accurately quantify driver risk perception, system intervention timing, and driver-vehicle collaboration patterns. Safety testing of autonomous vehicles currently relies primarily on real-world vehicle testing, but this approach faces numerous challenges, including high testing costs, low urgency, and difficulty in reproducing the data. This approach struggles to meet the demands for systematic research on high-risk scenarios for future intelligent driving technologies.

[0063] Based on this, this application proposes a safety testing method for autonomous driving vehicles that can collect human-vehicle interaction characteristics in dangerous scenarios.

[0064] The safety testing method for an autonomous driving vehicle provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the data acquisition device 102 communicates with the server 104, and the driving simulator 103 communicates with the server 104. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The server 104 instructs the driving simulator 103 to simulate the scenario corresponding to the multi-level test plan, and in the simulated scenario, the interaction information between the driver and the tested vehicle during the test is collected through the data acquisition device 102, and sent to the server 104, so that the server 104 can construct a multi-level autonomous driving human-vehicle interaction response data set based on the test information and the basic information of the driver. Among them, the server 104 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0065] In one embodiment, Figure 2 As shown, a safety test method for an autonomous vehicle is provided, which is applied to Figure 1 The following is an example of a server in the example, including:

[0066] S201, determining a multi-level test plan corresponding to the autonomous driving classification standard, wherein at least some of the multi-level test plans include: test content under dangerous scenarios.

[0067] Optionally, the autonomous driving classification standard can be determined based on the "Classification and Definition of Standard Road Vehicle Driving Automation Systems" of the Society of Automotive Engineers (SAE) or the "Automotive Driving Automation Classification" of the Ministry of Industry and Information Technology. Among them, the "Classification and Definition of Standard Road Vehicle Driving Automation Systems" divides autonomous driving technology into six levels, L0-L5. L0 means no automation, and the throttle, brake, and steering wheel are all controlled by the driver. It is the most common driving method. The driver can get assistance from warning and protection systems during driving; L1 is assisted driving, and the driving system provides a single assisted driving function, such as adaptive cruise control speed or lane keeping system assisted steering. The driver still needs to participate in and dominate the entire driving process; L2 is partial autonomous driving, and the vehicle can simultaneously autonomously control the vehicle's direction (lateral) and acceleration and braking (longitudinal), and realize automatic lane changing, automatic parking, automatic following and other functions. The driver can selectively divert his attention for some operations completed by the vehicle. However, the driver still needs to continuously monitor and be prepared to take over the vehicle; L3 is conditional autonomous driving. When the driving system meets the design operating conditions, It can perform all dynamic driving tasks, including environmental perception, steering and acceleration and deceleration control. When the designed operating conditions are about to be unmet, the driving system will request the driver to take over. Once the takeover request is issued, the driver needs to take over the vehicle immediately, that is, in specific scenarios, such as highways or morning and evening rush hours, the vehicle can be operated fully automatically; L4 is highly automated driving. In a specific area, all operations of the entire driving process are completed by the driving system alone. The driver does not assume the driving task and can completely divert his attention, but the driver needs to supervise the driving behavior of the automated driving system and can actively choose to take over the vehicle. If it exceeds the above-mentioned specific area, the driver may need to take over control of the vehicle; L5 is fully automated driving. The driving system can perform all driving tasks completely autonomously in any geographical area and road conditions, and at any time. No human intervention is required, and the vehicle can operate in a completely unmanned state. In the "Autonomous Driving Automation Classification", autonomous driving technology is also divided into six levels according to the degree of automation. L0 is emergency assistance, and the driving system cannot continuously perform the vehicle's lateral or longitudinal motion control in dynamic driving tasks, but has the ability to continuously perform some targets and events detection and response in dynamic driving tasks; L1 is partial driving assistance, and the driving system continuously performs the vehicle's lateral or longitudinal motion control in dynamic driving tasks under its designed operating conditions (ODD), and has the ability to detect and respond to some targets and events that are compatible with the performed vehicle lateral or longitudinal motion control.The driver and the automated driving system must jointly perform the driving task, and the driver must monitor the behavior of the automated driving system and perform appropriate responses or operations. Level 2 is combined driving assistance, in which the driving system continuously performs the vehicle's lateral and longitudinal motion control in dynamic driving tasks under designed operating conditions and has the ability to detect and respond to certain targets and events that are appropriate for the vehicle's lateral and longitudinal motion control being performed. Level 3 is conditional automated driving, which can continuously perform all dynamic driving tasks under designed operating conditions. When the automated driving system in operation issues a takeover request or in the event of a vehicle failure, a dynamic driving task backup user can take over. Level 4 is highly automated driving, which can continuously perform all dynamic driving tasks under designed operating conditions. When the automated driving system is unable to continue the dynamic driving task, the system will take measures to reduce the vehicle's accident risk to an acceptable level. Level 5 is fully automated driving, which can continuously perform all dynamic driving tasks under all drivable conditions. When the automated driving system is unable to continue the dynamic driving task, the system will take measures to reduce the vehicle's accident risk to an acceptable level.

[0068] It should be noted that the autonomous driving classification standard can also be any other classification standard.

[0069] In the embodiments of this application, a multi-level testing plan corresponding to the autonomous driving classification standard is determined based on the autonomous driving classification standard. For example, corresponding critical and non-critical scenarios can be designed for each level of autonomous driving, along with test content for the corresponding scenario. For example, the test content can include the type of critical situation and the type of data to be collected.

[0070] S202, simulating scenarios corresponding to multi-level test schemes through a driving simulator and obtaining test information corresponding to the multiple levels; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

[0071] The interactive information may include the driver's decision-making responses and vehicle dynamics data in different scenarios; or, the interactive information may include the driver's behavioral characteristics and interaction patterns obtained based on the driver's decision-making responses and vehicle dynamics data.

[0072] In an embodiment of the present application, the data acquisition device includes a vehicle acquisition device and a physiological signal measuring device. The vehicle acquisition device is connected to a driving simulator or is installed on a driving simulator, and the driver wears the required physiological signal measuring device during the test. Furthermore, the driving simulator simulates a scenario corresponding to at least one level of test scheme, and the interaction information between the driver and the tested vehicle during the test is collected through the data acquisition device. In an embodiment of the present application, the interaction information may include interaction information during the test in a dangerous scenario, as well as interaction information during the test in a non-dangerous scenario.

[0073] Optionally, driver characteristic information, such as height and weight, can be acquired in advance to determine the corresponding cab environment based on the driver's characteristic information. The cab environment may include steering wheel position, seat position, etc. The driver's eye distance from the screen and eye relief during normal driving can also be measured, allowing the parameters of the physiological signal measurement device to be adjusted based on this information.

[0074] As an optional implementation, for each driver, the server determines at least one level of test scheme from the multi-level test scheme as the target test scheme, and sends the target test scheme to the driving simulator so that the driving simulator can simulate the corresponding scenario according to the target test scheme.

[0075] As another optional implementation, the server can send all multi-level test plans to the driving simulator. The driving simulator randomly or according to the test requirements determines at least one level of test plan from the multi-level test plans as the target test plan, and simulates the corresponding scenario according to the target test plan.

[0076] In the above-mentioned application embodiment, a multi-level test scheme corresponding to the autonomous driving classification standard is determined, and at least some of the multi-level test schemes include: test content under dangerous scenarios; scenarios corresponding to the multi-level test schemes are simulated through a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios. By simulating scenarios corresponding to different levels of test schemes through a driving simulator, control and switching of different levels of autonomous driving can be achieved, and dynamic human-vehicle interaction of autonomous driving vehicles in dangerous scenarios can be carried out while ensuring the safety of the subjects. Efficiency is improved through dangerous acceleration testing, and the collection of human-vehicle interaction characteristics in dangerous scenarios is achieved.

[0077] In one embodiment, an implementation of the above S202 is provided, wherein at least one level test scheme in the multi-level test scheme includes: test content in non-dangerous scenarios, such as Figure 3As shown, the above-mentioned "simulating a multi-level test plan through a driving simulator to obtain test information corresponding to multiple levels" includes:

[0078] S301, simulating non-dangerous scenarios and dangerous scenarios through a driving simulator.

[0079] Dangerous scenarios include rear-end collisions, lane changes, and merging. Each level of testing can include multiple dangerous scenarios, and each dangerous scenario includes at least one dangerous scenario. Non-dangerous scenarios involve continuous traffic flow interaction around the tested vehicle, but the vehicle does not actively engage in dangerous behavior to create dangerous conditions.

[0080] In the embodiment of the present application, non-dangerous scenarios and dangerous scenarios may appear alternately, or the driver may complete the driving task in the dangerous scenario after completing the driving task in the non-dangerous scenario.

[0081] Optionally, the test process of the embodiment of the present application may include free driving and test driving. The free driving is used to make the driver familiar with the dynamic characteristics and interactive stimulation mechanism of the driving simulator. The free driving scenario is a non-dangerous scenario, and the data collected during the free driving process can be used as vehicle dynamics data in a non-dangerous scenario, as a benchmark reference for test driving and data analysis.

[0082] S302 , in non-dangerous scenarios and dangerous scenarios, collecting the driver's driving status information and vehicle status information to obtain multi-level corresponding test information.

[0083] In an embodiment of the present application, in a non-dangerous scenario, the driver's driving state information and vehicle state information can be collected for a preset time period after the driver simulates driving for a first time period. In a dangerous scenario, the driver's driving state information and vehicle state information can be collected for a preset time period after the driver simulates driving for a second time period. Optionally, the non-dangerous scenario can correspond to any level of the multi-level test scheme, or the non-dangerous scenario can correspond to the manual driving test scheme in the multi-level test scheme.

[0084] For example, the driver can first simulate driving in an urban scene with a speed limit of 90km / h. During this process, the driving simulation is purely manual driving. Traffic flow is constantly generated around the car to interact, but other cars will not actively generate dangerous behaviors to cause dangerous conditions. The driver is required to drive in this scene for 15 minutes. The driving task is to become familiar with the motion feedback of the driving simulator, be able to obey traffic rules and drive the vehicle skillfully and avoid collisions with other cars. Since it was determined in the preliminary experiment that most drivers can use the driving simulator proficiently within 5 minutes, it can be considered that the time after 5 minutes is the effective response of the subjects under normal driving conditions. Therefore, the data recorded during the 5-15 minutes of free driving can be used as the baseline data under normal driving conditions of the subjects in the dataset.

[0085] In the above-mentioned application embodiment, non-dangerous scenarios and dangerous scenarios are simulated by a driving simulator, so that the human-vehicle interaction characteristics in non-dangerous scenarios and the human-vehicle interaction characteristics in dangerous scenarios can be collected, making the collected data more comprehensive, and the environmental simulation through the driving simulator ensures the reliability of the data.

[0086] In one embodiment, an implementation of the above S301 is provided, such as Figure 4 As shown, the multi-level testing scheme includes: manual driving test scheme and multi-level autonomous driving test scheme. The above-mentioned "simulating dangerous scenarios through driving simulators" includes:

[0087] S401, simulating a dangerous scenario in a manual driving test plan through a driving simulator.

[0088] Among them, manual driving means non-automated driving.

[0089] In this embodiment of the present application, the manual driving test scenario corresponds to Level 0 autonomous driving. The server can first demonstrate to the driver the vehicle's functions and driver responsibilities for Level 0 simulated driving, then simulate a dangerous scenario according to the manual driving test scenario to conduct a Level 0 autonomous driving simulation test. Optionally, the test can last up to 30 minutes, with the driver required to drive in a highly realistic and immersive driving environment. If a collision occurs, the test is paused, a safety education video featuring a real-life collision incident is played to the driver, and the test then continues.

[0090] Optionally, dangerous scenarios corresponding to each level can be generated according to a preset dangerous scenario data generation method and a multi-level testing scheme.

[0091] S402, simulating a dangerous scenario in any level of autonomous driving test plan through a driving simulator.

[0092] In an embodiment of the present application, the autonomous driving level corresponding to the autonomous driving test plan can be L1 to L4. The server randomly determines the target driving level from multiple levels and determines the autonomous driving test plan corresponding to the target driving level as the target test plan. Furthermore, the driver can first be shown the car functions and driver responsibilities for simulated driving at the target driving level, and then simulate dangerous scenarios according to the manual driving test plan to conduct a simulated driving test for the target driving level. Optionally, the test process can last 30 minutes. The driver is required to drive in a highly realistic and immersive driving environment. If a collision occurs, the experiment is paused, and a safety education video containing a real collision accident is played to the driver before the test continues.

[0093] It's important to note that prior to the test, drivers only received a basic introduction to the vehicle's driving functions and operational experience, but were not given a specific level of automated driving. This prevents drivers from forming preconceptions about different levels of automated driving, which could affect their natural driving behavior. By comparing different groups of test subjects and quantifying the impact of different levels of automated driving on driving behavior and safety, we can ensure the universality and generalizability of the test results.

[0094] Optionally, after completing the test corresponding to the manual driving test plan, the driver can take a preset rest period before conducting the test corresponding to the autonomous driving test plan.

[0095] Optionally, in an embodiment of the present application, when the server randomly determines the target driving level from multiple levels, it is necessary to ensure that the tests at different levels are evenly distributed to avoid systematic errors, thereby making the data more statistically significant.

[0096] Optionally, the driver can determine whether to intervene or take over based on the driver's responsibilities and the status of the vehicle being tested. Once the driver intervenes, the control of the vehicle is completely handed over to the driver. Driver intervention may include steering or pedal braking.

[0097] Optionally, the total test duration may not exceed a preset test duration, such as 1.5 hours, to avoid increased driver fatigue and enhanced learning effects, which may affect data quality.

[0098] Optionally, for the same driver, multiple dangerous scenarios in manual driving test scenarios and dangerous scenarios in autonomous driving test scenarios can be generated.

[0099] In the above application embodiment, manual driving and any level of autonomous driving are used as test controls to control the driver's learning effect and fatigue impact, and the impact of different levels of autonomous driving on driving behavior and safety can be quantified, thereby ensuring the universality and generalizability of the test results.

[0100] In one embodiment, Figure 5 As shown, the above-mentioned safety testing method for the autonomous driving vehicle further includes:

[0101] S203: Construct a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the driver's basic information.

[0102] The basic information of the driver may include the basic background of the driver, judgment of driving behavior patterns, accident tendencies and driving style.

[0103] In an embodiment of the present application, a basic information input interface can be displayed to the driver. The input interface can specifically include input windows for a driver basic information scale, a driver behavior questionnaire (DBQ), and a multidimensional driving style inventory (MDSI). Based on the information input by the driver, the driver's basic background can be determined, and driving behavior patterns, accident propensity, and driving style can be judged. For example, the driver's basic background can be obtained based on the driver basic information scale, the driving behavior pattern and accident propensity can be determined by analyzing the driving behavior scale, and the driving style can be determined by analyzing the multidimensional driving style scale.

[0104] In an embodiment of the present application, test information is classified according to the level of autonomous driving to form data records such as vehicle dynamics, physiological signals, etc. of drivers driving autonomous vehicles of different levels facing dangerous interactions. Furthermore, result information such as collision rate and collision angle are extracted based on the data records, and then a human-vehicle interaction response data set of autonomous vehicles of different levels is constructed based on the result information.

[0105] Optionally, before constructing a multi-level autonomous driving human-vehicle interaction response dataset, the reliability of the simulated autonomous driving data can be verified. For example, test information collected using a driving simulator can be collated, including the driver's driving behavior data. This driving behavior data can be compared with natural driving data collected in the real physical world. Based on the probability distribution of driving behavior, such as speed distribution, acceleration distribution, and TTC distribution, metrics such as KL divergence (Kullback-Leibler divergence), JS divergence (Jensen-Shannon divergence), and W-distance (Wasserstein distance) can be used to determine the distribution difference between the driving behavior data and the natural driving data, thereby statistically demonstrating the similarity between the two. After the driver completes the driving simulator experiment, subjective ratings of the driver based on their experience at different levels of autonomous driving can be obtained, including but not limited to the realism of the first-person perspective presentation, the realism of the vehicle kinematic feedback, the realism of the surrounding vehicle kinematic behavior, the realism of the traffic scene, and the realism of the overall driving experience. The ratings can range from 0 to 10, with intervals of 1 point.

[0106] In the above-mentioned application embodiment, the test information obtained from the test and the basic information of the driver are combined to construct a multi-level autonomous driving human-vehicle interaction response data set, making the data in the data set more comprehensive and reliable.

[0107] In one embodiment, after constructing a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the basic information of the driver, Figure 6 As shown, the above-mentioned safety testing method for the autonomous driving vehicle further includes:

[0108] S204: Determine a test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset.

[0109] The autonomous driving human-vehicle interaction response dataset may include the number of tests for each critical scenario and the number of collisions in each critical scenario. The types of critical scenarios may include rear-end collisions, lane changes, and lane merges.

[0110] In the embodiment of the present application, for each dangerous situation type in each driving level, the ratio of the number of collisions to the number of tests can be determined as the test collision rate of the dangerous situation type.

[0111] S205 , obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver.

[0112] In the embodiments of the present application, due to differences in perception, decision-making, and control capabilities among autonomous vehicles of different driving levels, when faced with the same distribution of hazardous traffic scenarios, the collision outcomes and safety benefits of autonomous vehicles of different driving levels exhibit significant level-dependent differences. In order to quantify the collision risk of autonomous driving systems of different levels without driver intervention, this application introduces a baseline collision rate (BCR), which is the collision rate when decisions are made solely based on autonomous driving technology and the driver does not actively intervene. The baseline collision rate can reflect the inherent safety performance of autonomous driving systems of various levels in different hazardous scenarios and can be used to compare and analyze the adaptability and limitations of autonomous driving systems in complex traffic environments.

[0113] For example, based on algorithm simulation tests, the statistical results of baseline collision rates for autonomous vehicles at levels L1 to L4 in traffic flow are shown in Table 1. The data in Table 1 can be used to further analyze optimization directions for autonomous driving functions, evaluate the impact of driver intervention on collision rates, and explore the potential for safety improvements for different levels of autonomous driving systems in critical scenarios.

[0114] Table 1

[0115]

[0116] S206: Determine whether the vehicle under test meets the safety requirements of autonomous driving based on the baseline collision rate and the test collision rate.

[0117] In the embodiment of the present application, based on the baseline collision rate and the test collision rate, the impact of driver intervention on the collision rate can be evaluated, and the safety improvement potential of different levels of autonomous driving systems in dangerous scenarios can be explored.

[0118] As an optional implementation, the baseline collision rate and the test collision rate can be compared to determine whether driver intervention can reduce the collision rate. For example, for each autonomous driving level, if the baseline collision rate is lower than the test collision rate, the vehicle under test meets the autonomous driving safety requirements for that level.

[0119] Optionally, the baseline collision rate and the test collision rate correspond to the target dangerous state type, such as Figure 7 As shown, the above “determining whether the tested vehicle meets the safety requirements for autonomous driving based on the baseline collision rate and the test collision rate” includes:

[0120] S501: When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, it is determined that the autonomous driving safety requirements are not met.

[0121] S502: Based on the difference, the automatic driving behavior of the tested vehicle under the target dangerous state type is optimized and controlled.

[0122] In an embodiment of the present application, the target hazard type can be any of rear-end collision, lane change, and lane merging. For each target hazard type at each autonomous driving level, the degree of impact of driver intervention on the collision rate can be determined based on empirical values ​​or a prediction model, that is, the predicted difference range between the baseline collision rate and the test collision rate can be determined. The difference between the current test collision rate and the baseline collision rate is determined, and further, whether the difference between the baseline collision rate and the test collision rate is within the predicted difference range is determined. If so, it indicates that the vehicle under test meets the autonomous driving safety requirements at that autonomous driving level. If not, it indicates that the vehicle under test does not meet the autonomous driving safety requirements at that autonomous driving level.

[0123] Furthermore, if the vehicle under test does not meet the autonomous driving safety requirements at that autonomous driving level, the autonomous driving function corresponding to the target dangerous situation type of the vehicle under test is optimized and retested until the difference between the baseline collision rate and the test collision rate falls within the predicted difference range. Optionally, a correspondence between the parameter adjustment scheme and the difference can be pre-set, so that the parameters of the autonomous driving function corresponding to the target dangerous situation type are adjusted based on the correspondence and the difference.

[0124] In the above application embodiment, the baseline collision rate and the test collision rate can be used to further analyze the optimization direction of the autonomous driving function, evaluate the impact of driver intervention on the collision rate, and evaluate the safety improvement potential of different levels of autonomous driving systems in dangerous scenarios.

[0125] In one embodiment, after constructing a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the basic information of the driver, Figure 8 As shown, the above-mentioned safety testing method for the autonomous driving vehicle further includes:

[0126] S207 , obtaining safety assessment parameters from the response data set, where the safety assessment parameters include at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter.

[0127] Among them, collision risk parameters can measure the possibility of collision of autonomous vehicles in specific scenarios. Collision risk parameters can include collision probability, collision energy, vehicle spacing and traffic density; vehicle dynamic parameters can analyze vehicle dynamic parameters such as acceleration and braking intensity. Vehicle dynamic parameters can include longitudinal acceleration, lateral acceleration, braking intensity and automatic intervention success rate; time margin parameters can determine the reaction time and decision margin of the driver or autonomous driving system. Time margin can include avoidance time margin, minimum time distance, driver reaction time, and autonomous driving control delay; trajectory control can evaluate the vehicle's control ability for lane keeping and trajectory planning. Trajectory control parameters can include vehicle path offset, lane departure angle visual range, intervention success rate, etc.

[0128] For example, the security assessment parameters may be as shown in Table 2:

[0129] Table 2

[0130]

[0131] S208: Determine a safety score of the tested vehicle based on the safety assessment parameters.

[0132] In this embodiment, the 16 key safety parameters in the Safety Benefit Evaluation Matrix (SBEM) shown in Table 2 were populated based on actual driving data and simulation test results to ensure that the safety performance reflected in the matrix is ​​based on real-world evidence. A fuzzy analytic hierarchy process (AHP) was used to determine weights for each evaluation category and its associated parameters. This method quantifies the relative importance of each parameter in the overall safety assessment, optimizes the reliability and consistency of the weight calculation, and determines the safety score of the tested vehicle.

[0133] Specifically, since autonomous driving safety assessment involves multiple uncertainties, the Fuzzy Analytic Hierarchy Process (FAHP) is used, and triangular fuzzy numbers (TFNs) are introduced to represent the importance of safety parameters, making the assessment more scientific and reasonable. The FAHP determination steps are as follows:

[0134] (1) Constructing a fuzzy judgment matrix: using triangular fuzzy numbers to compare the importance of parameters pairwise and reduce the uncertainty of subjective scoring;

[0135] (2) Calculate fuzzy weights: Use fuzzy synthesis method to calculate the fuzzy weights of each parameter and perform normalization to ensure that the weight distribution is reasonable;

[0136] (3) Defuzzification: Using methods such as the centroid method or the α-cut method, the final accurate weights are calculated to map them to the safety assessment matrix;

[0137] (4) Consistency check: Ensure that the calculated weights meet the consistency criteria and improve the reliability of decision making.

[0138] Furthermore, the safety score of the tested vehicle is determined based on the safety benefit evaluation matrix shown in Table 2 and the weight of each parameter.

[0139] In the above-mentioned application embodiment, a quantitative evaluation is performed on the overall safety benefits of the tested vehicle to provide data support for autonomous driving decision optimization, intelligent safety system design and personalized driving intervention strategies. The fuzzy hierarchical analysis method is used to optimize weight calculation, which can improve the accuracy and robustness of the evaluation results. It can be used for safety assessments of different levels of autonomous driving, and is suitable for actual road testing and simulation verification, thereby improving the safety, reliability and adaptability of the autonomous driving system.

[0140] In one embodiment, an implementation of the above S201 is provided, such as Figure 9 As shown in the figure, the above-mentioned “determining a multi-level test plan corresponding to the autonomous driving classification standard” includes:

[0141] S601: Determine the vehicle functions corresponding to each level of autonomous driving according to the autonomous driving classification standard.

[0142] In the embodiments of this application, the vehicle functions corresponding to autonomous driving include, but are not limited to, adaptive cruise control (ACC), automatic emergency braking (AEB), lane keeping control (LKA), and automatic emergency steering (AES). By embedding these driving assistance functions in a simulation environment, the vehicle can perceive risks and respond autonomously to varying degrees in complex traffic flows and dangerous traffic scenarios, thereby meeting the functional requirements of various levels of autonomous driving. Specifically, according to the autonomous driving classification standards, the vehicle functions corresponding to various levels of autonomous driving may include:

[0143] (1) L0 (no automation): Pure manual driving, the driver has full control over the vehicle, and all decisions are made by the driver; the driving system only provides warning auxiliary functions (such as collision warning, blind spot monitoring, etc.), but does not perform any automated intervention.

[0144] (2) L1 (driving assistance): It has a single control function of adaptive cruise control (ACC), which can autonomously adjust the vehicle speed and achieve longitudinal control; it has no lateral control capability, cannot autonomously avoid collisions or maintain lanes, and the driver needs to always monitor the vehicle's driving status.

[0145] (3) L2 (partial automation): The driving system can simultaneously perform longitudinal (acceleration and deceleration) and lateral (steering) control, and supports functions such as automatic emergency braking and lane keeping systems. It can only cope with limited dangerous traffic scenarios, such as triggering AEB when the vehicle in front brakes suddenly, but driver intervention is still required in complex interactive scenarios such as dynamic collision avoidance. The driver still needs to maintain supervision of the driving process and be able to take over at any time to deal with situations that the system cannot handle.

[0146] (4) L3 (conditional automation): The vehicle can drive autonomously in a specific environment (such as a highway) and can autonomously respond to complex traffic conditions within a certain range, such as automatic emergency steering. The driving system can request the driver to take over in some situations (such as when the system detects a situation beyond its handling range). If the driving system issues a takeover reminder or the driver determines that the system is about to fail, the driver must intervene in a timely manner, otherwise driving safety may be affected.

[0147] (5) L4 (high automation): The vehicle can achieve fully autonomous driving in limited scenarios such as closed campuses or highways, and no driver intervention is required under normal circumstances; the driving system can independently respond to most complex traffic conditions and dangerous scenarios, including dynamic collision avoidance and emergency braking; in a few extreme dangerous scenarios, the system may fail to correctly identify the risk and will not actively issue a takeover request, but a collision may occur. Among them, extreme dangerous scenarios include exceeding the system design domain and the system being unable to predict collision avoidance failure.

[0148] S602: Determine a multi-level test plan based on the vehicle functions corresponding to each level of autonomous driving.

[0149] In the embodiment of this application, three typical types of dangerous situations are selected to fully cover the sudden risks that autonomous vehicles may face in highway environments:

[0150] (1) Rear-end collision: When the vehicle ahead suddenly brakes or slows down, the vehicle needs to quickly sense and respond urgently (such as braking or changing lanes).

[0151] (2) Lane change: A vehicle in an adjacent lane suddenly cuts in front of the vehicle, causing the distance between vehicles to shorten rapidly, which tests the response strategies of the autonomous driving system and the driver.

[0152] (3) Merge: When the vehicle is changing lanes, if a sudden obstacle or a rapidly approaching vehicle appears in the target lane, a quick decision on a collision avoidance plan is required.

[0153] In the embodiments of the present application, for each level of autonomous driving, the dangerous trigger mechanism is optimized through time range control, trigger strategy optimization, and urgency regulation. In order to enhance the randomness and authenticity of the test, the trigger time of dangerous events is dynamically adjusted within the range of 60s±20s to ensure a balanced distribution of different dangerous events during the test; the triggering time of dangerous scenarios adopts a random selection mechanism, and the server randomly selects a dangerous type at a preset time point or a random time point to avoid the driver from forming a fixed pattern of prejudgment behavior, thereby enhancing the effectiveness of the test; the urgency of dangerous scenarios conforms to the normal distribution, that is, most triggering events are at a moderate level of urgency, but still include some extreme dangerous situations with a high level of urgency, ensuring the wide applicability and reliability of the experimental results.

[0154] In the above-mentioned application embodiment, in the determined multi-level test scheme, through the random triggering and accelerated generation of dangerous events, the driver can experience high-frequency and high-intensity dangerous scenarios in a short period of time, thereby improving data collection efficiency; and because the triggering timing and type of dangerous events are unpredictable, it is difficult for the driver to adapt or adjust the driving behavior in advance, reducing the learning and prediction effects, thereby ensuring the authenticity and representativeness of the data.

[0155] In one embodiment, a complete safety testing method for an autonomous vehicle is provided, such as Figure 10 As shown, the above method includes:

[0156] S1, according to the autonomous driving classification standards, determine the vehicle functions corresponding to each level of autonomous driving.

[0157] S2. Determine a multi-level test plan based on the vehicle functions corresponding to each level of autonomous driving; at least some of the multi-level test plans include: test content under dangerous scenarios.

[0158] S3, simulating a non-dangerous scenario through a driving simulator, collecting the driver's driving status information and vehicle status information to obtain test information corresponding to the non-dangerous scenario.

[0159] S4, simulate dangerous scenarios in manual driving test scenarios through driving simulators, and simulate dangerous scenarios in any level of autonomous driving test scenarios through driving simulators.

[0160] S5, collecting the driver's driving status information and vehicle status information to obtain multi-level corresponding test information.

[0161] S6, based on the test information and the basic information of the driver, constructs a multi-level autonomous driving human-vehicle interaction response dataset.

[0162] S7, determining the test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset.

[0163] S8, obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver.

[0164] S9: When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, it is determined that the autonomous driving safety requirements are not met.

[0165] S10, based on the difference, optimizing and controlling the autonomous driving behavior of the tested vehicle under the target dangerous state type.

[0166] S11, obtaining safety assessment parameters from the response data set, where the safety assessment parameters include at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter.

[0167] S12, determining a safety score of the tested vehicle based on the safety assessment parameters.

[0168] In the above-mentioned application embodiment, a multi-level test scheme corresponding to the autonomous driving classification standard is determined, and at least some of the multi-level test schemes include: test content under dangerous scenarios; scenarios corresponding to the multi-level test schemes are simulated through a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios. By simulating scenarios corresponding to different levels of test schemes through a driving simulator, control and switching of different levels of autonomous driving can be achieved, and dynamic human-vehicle interaction of autonomous driving vehicles in dangerous scenarios can be carried out while ensuring the safety of the subjects. Efficiency is improved through dangerous acceleration testing, and the collection of human-vehicle interaction characteristics in dangerous scenarios is achieved.

[0169] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0170] Based on the same inventive concept, embodiments of the present application also provide a safety testing device for an autonomous vehicle for implementing the aforementioned safety testing method for an autonomous vehicle. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more of the following embodiments of the safety testing device for an autonomous vehicle can be found in the aforementioned limitations of the safety testing method for an autonomous vehicle, and will not be further elaborated here.

[0171] In one embodiment, Figure 11 As shown, a safety testing device for an autonomous driving vehicle is provided, comprising: a first determining module 10, a first acquiring module 11 and a constructing module 12, wherein:

[0172] The first determination module 10 is used to determine a multi-level test plan corresponding to the autonomous driving classification standard, where at least some of the multi-level test plans include: test content under dangerous scenarios.

[0173] The first acquisition module 11 is used to simulate the scenarios corresponding to the multi-level test scheme through a driving simulator and obtain test information corresponding to the multiple levels; the test information includes: interaction information between the driver and the tested vehicle during the test process.

[0174] In one embodiment, the acquisition module 11 includes: a simulation unit and a collection unit, wherein:

[0175] The simulation unit is used to simulate non-dangerous scenarios and dangerous scenarios through a driving simulator.

[0176] The collection unit is used to collect the driver's driving status information and vehicle status information in non-dangerous scenarios and dangerous scenarios to obtain multi-level corresponding test information.

[0177] In one embodiment, the above-mentioned simulation unit is specifically used to simulate dangerous scenarios in manual driving test scenarios through a driving simulator; and to simulate dangerous scenarios in any level of autonomous driving test scenarios through a driving simulator.

[0178] In one embodiment, the safety testing device for the autonomous driving vehicle further includes:

[0179] Construction module 12 is used to construct a multi-level autonomous driving human-vehicle interaction response dataset based on test information and basic driver information.

[0180] In one embodiment, the safety testing device for the autonomous driving vehicle further includes: a second determination module, a second acquisition module, and a third determination module, wherein:

[0181] The second determination module is used to determine the test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response data set.

[0182] The second acquisition module is used to obtain a reference collision rate of the vehicle under test, where the reference collision rate is the collision rate of the vehicle under test without a driver.

[0183] The third determination module is used to determine whether the vehicle under test meets the autonomous driving safety requirements based on the baseline collision rate and the test collision rate.

[0184] In one embodiment, the third determination module includes: a first determination unit and an optimization unit, wherein:

[0185] The first determination unit is used to determine that the autonomous driving safety requirement is not met when the difference between the baseline collision rate and the test collision rate is not within the predicted difference range.

[0186] The optimization unit is used to optimize the automatic driving behavior of the tested vehicle under the target dangerous state type based on the difference.

[0187] In one embodiment, the safety testing device for the autonomous driving vehicle further includes: a second acquisition module and a fourth determination module, wherein:

[0188] The second acquisition module is used to obtain safety assessment parameters from the response data set, where the safety assessment parameters include at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter.

[0189] The fourth determination module is used to determine the safety score of the tested vehicle based on the safety assessment parameters.

[0190] In one embodiment, the first determining module includes: a second determining unit and a third determining unit, wherein:

[0191] The second determination unit is used to determine the vehicle functions corresponding to each level of autonomous driving according to the autonomous driving classification standard.

[0192] The third determination unit is used to determine a multi-level test plan according to the vehicle functions corresponding to each level of autonomous driving.

[0193] Each module in the aforementioned autonomous vehicle safety testing device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0194] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 12 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store safety test data of the autonomous driving vehicle. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a safety testing method for an autonomous driving vehicle is implemented.

[0195] Those skilled in the art will understand that Figure 12 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0196] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0197] Determine a multi-level test plan corresponding to the automated driving classification standard, where at least some of the multi-level test plans include: testing content for hazardous scenarios;

[0198] The driving simulator is used to simulate the scenarios corresponding to the multi-level test scheme and obtain the test information corresponding to the multi-level test scheme; the test information includes: the interaction information between the driver and the vehicle under test during the test process, and the interaction information at least includes the interaction information during the test process under dangerous scenarios.

[0199] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0200] Simulate non-critical and critical scenarios through driving simulators;

[0201] In non-dangerous scenarios and dangerous scenarios, the driver's driving status information and vehicle status information are collected to obtain multi-level corresponding test information.

[0202] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0203] Simulate dangerous scenarios in manual driving test scenarios through driving simulators;

[0204] Use the driving simulator to simulate dangerous scenarios in any level of autonomous driving test scenario.

[0205] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0206] Based on the test information and the basic information of the driver, a multi-level autonomous driving human-vehicle interaction response dataset is constructed.

[0207] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0208] Determine the test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset;

[0209] Obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver;

[0210] Based on the baseline collision rate and the test collision rate, determine whether the vehicle under test meets the safety requirements for autonomous driving.

[0211] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0212] When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, it is determined that the autonomous driving safety requirements are not met;

[0213] Based on the difference, the autonomous driving behavior of the tested vehicle under the target dangerous state type is optimized and controlled.

[0214] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0215] Obtaining safety assessment parameters from the response data set, the safety assessment parameters including at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter;

[0216] Based on the safety assessment parameters, a safety score of the tested vehicle is determined.

[0217] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0218] Determine the vehicle functions corresponding to each level of autonomous driving based on the autonomous driving classification standards;

[0219] Determine multi-level testing plans based on the vehicle functions corresponding to each level of autonomous driving.

[0220] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0221] Determine a multi-level test plan corresponding to the automated driving classification standard, where at least some of the multi-level test plans include: testing content for hazardous scenarios;

[0222] The driving simulator is used to simulate the scenarios corresponding to the multi-level test scheme and obtain the test information corresponding to the multi-level test scheme; the test information includes: the interaction information between the driver and the vehicle under test during the test process, and the interaction information at least includes the interaction information during the test process under dangerous scenarios.

[0223] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0224] Simulate non-critical and critical scenarios through driving simulators;

[0225] In non-dangerous scenarios and dangerous scenarios, the driver's driving status information and vehicle status information are collected to obtain multi-level corresponding test information.

[0226] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0227] Simulate dangerous scenarios in manual driving test scenarios through driving simulators;

[0228] Use the driving simulator to simulate dangerous scenarios in any level of autonomous driving test scenario.

[0229] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0230] Based on the test information and the basic information of the driver, a multi-level autonomous driving human-vehicle interaction response dataset is constructed.

[0231] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0232] Determine the test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset;

[0233] Obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver;

[0234] Based on the baseline collision rate and the test collision rate, determine whether the vehicle under test meets the safety requirements for autonomous driving.

[0235] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0236] When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, it is determined that the autonomous driving safety requirements are not met;

[0237] Based on the difference, the autonomous driving behavior of the tested vehicle under the target dangerous state type is optimized and controlled.

[0238] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0239] Obtaining safety assessment parameters from the response data set, the safety assessment parameters including at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter;

[0240] Based on the safety assessment parameters, a safety score of the tested vehicle is determined.

[0241] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0242] Determine the vehicle functions corresponding to each level of autonomous driving based on the autonomous driving classification standards;

[0243] Determine multi-level testing plans based on the vehicle functions corresponding to each level of autonomous driving.

[0244] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0245] Determine a multi-level test plan corresponding to the automated driving classification standard, where at least some of the multi-level test plans include: testing content for hazardous scenarios;

[0246] The driving simulator is used to simulate the scenarios corresponding to the multi-level test scheme and obtain the test information corresponding to the multi-level test scheme; the test information includes: the interaction information between the driver and the vehicle under test during the test process, and the interaction information at least includes the interaction information during the test process under dangerous scenarios.

[0247] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0248] Simulate non-critical and critical scenarios through driving simulators;

[0249] In non-dangerous scenarios and dangerous scenarios, the driver's driving status information and vehicle status information are collected to obtain multi-level corresponding test information.

[0250] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0251] Simulate dangerous scenarios in manual driving test scenarios through driving simulators;

[0252] Use the driving simulator to simulate dangerous scenarios in any level of autonomous driving test scenario.

[0253] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0254] Based on the test information and the basic information of the driver, a multi-level autonomous driving human-vehicle interaction response dataset is constructed.

[0255] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0256] Determine the test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset;

[0257] Obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver;

[0258] Based on the baseline collision rate and the test collision rate, determine whether the vehicle under test meets the safety requirements for autonomous driving.

[0259] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0260] When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, it is determined that the autonomous driving safety requirements are not met;

[0261] Based on the difference, the autonomous driving behavior of the tested vehicle under the target dangerous state type is optimized and controlled.

[0262] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0263] Obtaining safety assessment parameters from the response data set, the safety assessment parameters including at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter;

[0264] Based on the safety assessment parameters, a safety score of the tested vehicle is determined.

[0265] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0266] Determine the vehicle functions corresponding to each level of autonomous driving based on the autonomous driving classification standards;

[0267] Determine multi-level testing plans based on the vehicle functions corresponding to each level of autonomous driving.

[0268] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0269] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0270] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A safety testing method for an autonomous driving vehicle, characterized in that: The method comprises: Determining a multi-level test plan corresponding to the automated driving classification standard, wherein at least some of the multi-level test plans include: test content for dangerous scenarios; The scenarios corresponding to the multi-level test scheme are simulated by a driving simulator, and test information corresponding to the multiple levels is obtained; the test information includes: interaction information between the driver and the tested vehicle during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

2. The method according to claim 1, characterized in that At least one level test scheme in the multi-level test scheme includes: test content in a non-dangerous scenario; simulating the multi-level test scheme by a driving simulator to obtain test information corresponding to the multiple levels includes: simulating the non-dangerous scenario and the dangerous scenario by the driving simulator; In the non-dangerous scenario and the dangerous scenario, the driving state information and vehicle state information of the driver are collected to obtain test information corresponding to multiple levels.

3. The method according to claim 2, characterized in that The multi-level testing scheme includes: a manual driving test scheme and a multi-level autonomous driving test scheme. The simulating the dangerous scenario by a driving simulator includes: Simulating the dangerous scenario in the manual driving test scheme by the driving simulator; The driving simulator is used to simulate dangerous scenarios in any level of autonomous driving test scheme.

4. The method according to claim 1, wherein The method further comprises: A multi-level autonomous driving human-vehicle interaction response dataset is constructed based on the test information and the basic information of the driver.

5. The method according to claim 4, characterized in that After constructing a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the basic information of the driver, the method further includes: Determining a test collision rate of the vehicle under test based on the autonomous driving human-vehicle interaction response dataset; Obtaining a baseline collision rate of the vehicle under test, where the baseline collision rate is the collision rate of the vehicle under test without a driver; Based on the baseline collision rate and the test collision rate, determine whether the vehicle under test meets the autonomous driving safety requirements.

6. The method according to claim 5, characterized in that The baseline collision rate and the test collision rate correspond to a target dangerous state type; and determining whether the tested vehicle meets the autonomous driving safety requirements based on the baseline collision rate and the test collision rate includes: When the difference between the baseline collision rate and the test collision rate is not within the predicted difference range, determining that the autonomous driving safety requirement is not met; Based on the difference, the automatic driving behavior of the tested vehicle under the target dangerous state type is optimized and controlled.

7. The method according to claim 4, characterized in that After constructing a multi-level autonomous driving human-vehicle interaction response dataset based on the test information and the basic information of the driver, the method further includes: Obtaining safety assessment parameters from the response data set, the safety assessment parameters including at least one of a collision risk parameter, a vehicle dynamic parameter, a time margin parameter, and a trajectory control parameter; Based on the safety assessment parameters, a safety score of the tested vehicle is determined.

8. The method according to any one of claims 1 to 7, characterized in that Determining a multi-level testing plan corresponding to the autonomous driving classification standard includes: Determining the vehicle functions corresponding to each level of autonomous driving according to the autonomous driving classification standards; The multi-level testing scheme is determined according to the vehicle functions corresponding to each level of autonomous driving.

9. A safety testing device for an autonomous driving vehicle, characterized in that: The device comprises: A first determining module is configured to determine a multi-level test scheme corresponding to the autonomous driving classification standard, wherein at least some of the multi-level test schemes include: test content under dangerous scenarios; The first acquisition module is used to simulate the scenarios corresponding to the multi-level test scheme through a driving simulator and obtain test information corresponding to the multiple levels; the test information includes: interaction information between the driver and the vehicle under test during the test process, and the interaction information at least includes interaction information during the test process under dangerous scenarios.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images