Method and device for evaluating safety of automatic driving vehicle during whole driving process and computer equipment

Abstract

The application relates to a kind of automatic driving car driving whole process safety evaluation method, device and computer equipment, the method obtains the first driving data of target vehicle, and obtains the second driving data of vehicle in the surrounding area of target vehicle, then the first driving data and the second driving data are input into risk assessment model to carry out risk assessment to target vehicle, obtain risk assessment result, finally risk assessment result is input into safety assessment model to carry out safety assessment to target vehicle, obtain safety assessment result.Risk assessment result includes driving risk, collision risk and damage risk.In the above method, risk assessment model and safety assessment model are used, combined with driving risk, collision risk and damage risk, more comprehensive and more in-depth evaluation of the safety of automatic driving vehicle is realized, and the accuracy of the evaluation can be improved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a method, apparatus, and computer equipment for safety assessment of the entire driving process of an autonomous vehicle. Background Technology

[0002] As artificial intelligence matures, cities are becoming increasingly intelligent, and autonomous vehicles represent an emerging industry within the field of AI. While the continuous development of autonomous driving technology may gradually reduce its accident rate, existing data is insufficient to fully support safety comparisons between vehicles of different levels of autonomous driving in complex and critical scenarios.

[0003] Currently, safety assessments of autonomous vehicles primarily compare the safety of vehicles at different levels of automation using three main methods: First, target crash population analysis, which estimates the effectiveness of autonomous driving technology in preventing potential collisions by matching autonomous driving functions with specific accident types (rear-end collisions, "ghost pedestrian" scenarios). Second, traffic flow simulation, which assesses the impact of autonomous vehicles at different levels and penetration rates on overall traffic flow dynamics by constructing virtual traffic environments. Third, research based on driving simulators, which evaluates human-vehicle interaction responses in specific scenarios.

[0004] However, the safety assessment methods for autonomous vehicles suffer from inaccurate assessments. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, and computer equipment for assessing the safety of autonomous vehicles throughout their entire driving process, which can improve the accuracy of safety assessments during the entire driving process of autonomous vehicles, in response to the aforementioned technical problems.

[0006] Firstly, this application provides a method for safety assessment of an autonomous vehicle throughout its driving process, the method comprising:

[0007] Acquire the first driving data of the target vehicle, and acquire the second driving data of vehicles in the surrounding area of ​​the target vehicle;

[0008] The first and second driving data are input into the risk assessment model to conduct a risk assessment of the target vehicle and obtain the risk assessment results; the risk assessment results include driving risk, collision risk and damage risk.

[0009] The risk assessment results are input into the safety assessment model to conduct a safety assessment of the target vehicle, and the safety assessment results are obtained.

[0010] In some embodiments, the risk assessment model includes a driving risk assessment sub-model, a collision risk assessment sub-model, and a damage risk assessment sub-model. First driving data and second driving data are input into the risk assessment model to perform a risk assessment of the target vehicle, obtaining risk assessment results, including:

[0011] The initial driving data is input into the driving risk assessment sub-model for evaluation to obtain the driving risk.

[0012] The first and second driving data are input into the collision risk assessment sub-model for evaluation to obtain the collision risk.

[0013] The first and second driving data are input into the damage risk assessment sub-model for evaluation to obtain the damage risk.

[0014] In some embodiments, the driving risk assessment sub-model includes a weighting analysis sub-model and a first calculation sub-model. First driving data is input into the driving risk assessment sub-model for evaluation to obtain the driving risk, including:

[0015] The first driving data is input into the weight analysis sub-model to obtain the weights of the target factors; the target factors include vehicle speed, acceleration, lane departure distance, and environmental factors.

[0016] The weights of the target factors are input into the first calculation sub-model to obtain the driving risk.

[0017] In some embodiments, the collision risk assessment sub-model includes an analysis sub-model and a second calculation sub-model. First driving data and second driving data are input into the collision risk assessment sub-model for evaluation to obtain the collision risk, including:

[0018] Input the first driving data and the second driving data into the analysis sub-model to obtain the longitudinal collision risk and the lateral collision risk;

[0019] The longitudinal and lateral collision risks are input into the second calculation sub-model to obtain the collision risk.

[0020] In some embodiments, the method for obtaining the damage risk assessment sub-model includes:

[0021] Acquire collision speed sample data and other sample data of the vehicle under collision conditions; other sample data include at least one of the following: initial collision condition sample data, occupant physiological parameter sample data, and occupant posture sample data.

[0022] The other sample data is processed in high dimension to obtain the processed other sample data, and the initial damage risk assessment sub-model is trained based on the processed other sample data to obtain the damage risk assessment sub-model.

[0023] In some embodiments, the method further includes:

[0024] Obtain a safety assessment scale; the safety assessment scale includes multiple driving factors, each driving factor includes multiple driving safety questions, and the corresponding scores for each driving safety question;

[0025] A safety assessment model is constructed based on the safety assessment scale and driving risk parameters, collision risk parameters, and damage risk parameters.

[0026] In some embodiments, a safety assessment model is constructed based on a safety assessment scale and driving risk parameters, collision risk parameters, and damage risk parameters, including:

[0027] Construct a driving factor load matrix based on the safety assessment scale;

[0028] Determine the load value corresponding to each driving factor based on the driving factor load matrix;

[0029] The weights of driving risk parameters, collision risk parameters, and damage risk parameters are determined based on the load values ​​corresponding to each driving factor and the scores corresponding to each driving safety question.

[0030] A safety assessment model is constructed based on the weights of driving risk parameters, collision risk parameters, damage risk parameters, driving risk parameters, collision risk parameters, and damage risk parameters.

[0031] Secondly, this application also provides a safety assessment device for the entire driving process of an autonomous vehicle, the device comprising:

[0032] The acquisition module is used to acquire the first driving data of the target vehicle and the second driving data of vehicles in the surrounding area of ​​the target vehicle.

[0033] The risk assessment module is used to input first and second driving data into the risk assessment model to conduct a risk assessment of the target vehicle and obtain the risk assessment results; the risk assessment results include driving risk, collision risk and damage risk.

[0034] The safety assessment module is used to input the risk assessment results into the safety assessment model to conduct a safety assessment of the target vehicle and obtain the safety assessment results.

[0035] Thirdly, this application also provides a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method for safety assessment of the entire driving process of an autonomous vehicle as described in any of the embodiments of the first aspect above.

[0036] Fourthly, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method for safety assessment of the entire driving process of an autonomous vehicle as described in any of the embodiments of the first aspect above.

[0037] Fifthly, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the method for safety assessment of the entire driving process of an autonomous vehicle as described in any of the embodiments of the first aspect above.

[0038] The aforementioned method, apparatus, and computer equipment for safety assessment of the entire driving process of autonomous vehicles involve acquiring first driving data of the target vehicle and second driving data of vehicles in the surrounding area. This first and second driving data are then input into a risk assessment model to perform a risk assessment of the target vehicle, yielding a risk assessment result. Finally, the risk assessment result is input into a safety assessment model to perform a safety assessment of the target vehicle, yielding a safety assessment result. The risk assessment result includes driving risk, collision risk, and injury risk. This method utilizes the risk assessment model and the safety assessment model, combining driver behavior data and vehicle dynamics data from both normal and hazardous traffic scenarios. It extends safety assessment from hazardous scenarios forward to the normal driving process and backward to collisions and occupant injury, extending safety benefit assessment from traditional driving risk quantification to collision risk quantification and occupant injury risk assessment. This achieves a more comprehensive and in-depth assessment of the safety of autonomous vehicles, thereby improving the accuracy of the assessment. Attached Figure Description

[0039] Figure 1 These are internal structural diagrams of the computer device in some embodiments;

[0040] Figure 2 This is one of the flowcharts illustrating the safety assessment method for the entire driving process of an autonomous vehicle in some embodiments;

[0041] Figure 3 This is the second flowchart illustrating the safety assessment method for the entire driving process of an autonomous vehicle in some embodiments;

[0042] Figure 4 This is the third flowchart illustrating the safety assessment method for the entire driving process of an autonomous vehicle in some embodiments;

[0043] Figure 5 This is the fourth flowchart illustrating the safety assessment method for the entire driving process of an autonomous vehicle in some embodiments;

[0044] Figure 6This is the fifth flowchart illustrating the safety assessment method for the entire driving process of an autonomous vehicle in some embodiments;

[0045] Figure 7 This is the sixth flowchart illustrating the safety assessment method for the entire driving process of an autonomous vehicle in some embodiments;

[0046] Figure 8 This is a structural block diagram of a safety assessment device for the entire driving process of an autonomous vehicle, as shown in some embodiments. Detailed Implementation

[0047] In the embodiments of this application, the term "and / or" describes the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following associated objects have an "or" relationship.

[0048] In the embodiments of this application, the term "multiple" refers to two or more, and other quantifiers are similar.

[0049] In the embodiments of this application, the term "at least one" means one or more. For example, at least one of A, B and C can represent six situations: A exists alone, B exists alone, C exists alone, A and B exist simultaneously, A and C exist simultaneously, B and C exist simultaneously, and A, B and C exist simultaneously.

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0051] Road traffic accidents cause 1.19 million deaths globally each year, with occupant fatalities accounting for 29% of all road accident deaths, the largest proportion. The road safety situation is severe, and improving vehicle safety protection and reducing road traffic accidents and injuries are urgent social problems that need to be addressed. Vehicle safety and participant protection are pillar research areas for improving road traffic safety. Road traffic scenarios are complex and ever-changing, with the dynamic interaction of the "human-vehicle-road" triad forming a highly dynamic, strongly coupled, and nonlinear generalized dynamic system. Hazardous traffic scenarios, distinct from normal driving scenarios, refer to traffic scenarios where there are traffic conflict sources and the potential for a collision at any time. The hazardous scenario stage plays a crucial role in safety risk. When a hazardous traffic scenario occurs, the driver goes through three main processes: hazard perception, driving decision-making, and collision avoidance execution, controlling the vehicle's interaction with the hazard source and influencing whether a collision occurs. When a collision is not avoided, the human body experiences impact loads and suffers injury during the collision phase. As automotive intelligence levels gradually increase, autonomous vehicles are considered to significantly improve overall road safety due to their ability to significantly reduce traffic accidents caused by human driver error, distraction, or fatigue. However, existing data indicates that the actual safety performance of autonomous vehicles has not yet significantly surpassed that of traditional vehicles. According to statistics from the National Highway Traffic Safety Administration (NHTSA) in 2021, autonomous vehicles average 9.1 collisions per million miles driven, compared to 4.2 for traditional vehicles. This comparison highlights the safety shortcomings of autonomous driving technology at its current stage. Furthermore, assessing the safety of autonomous vehicles still faces numerous challenges. A core issue is the scarcity of safety-critical scenarios, which require large-scale analysis and verification of real-world accident data. However, existing real-world accident statistics have limitations in the following aspects: comparing the safety of vehicles at different levels of autonomy does not fully consider the relative consistency of road environment, accident type, vehicle state before the accident, and the urgency of the traffic scenario, resulting in some safety conclusions being impossible to directly compare due to highly inconsistent prior conditions. Furthermore, significant differences exist between vehicles of different levels of autonomous driving in terms of quantity, testing sites, operating conditions, and accident rate assessment criteria, making it difficult to compare their safety under similar or comparable conditions. Although the continuous development of autonomous driving technology may gradually reduce its accident rate, existing data is insufficient to fully support safety comparisons between vehicles of different levels of autonomous driving in complex and critical scenarios. Existing research mainly compares the safety of vehicles of different levels of autonomous driving using the following three methods: First, target crash population analysis, which estimates the effectiveness of autonomous driving technology in preventing potential collisions by matching autonomous driving functions with specific accident types (rear-end collisions, "ghost pedestrian" scenarios).Second, traffic flow simulation involves constructing virtual traffic environments to assess the impact of autonomous vehicles of different levels and penetration rates on overall traffic flow dynamics. Third, research based on driving simulators can fully consider driver behavior characteristics and evaluate human-vehicle interaction responses in real time under specific scenarios. Although existing methods are effective in some scenarios, they still have significant limitations in data collection and safety benefit comparison in hazardous scenarios. First, target collision population-based analysis methods are only applicable to low-level autonomous driving systems, such as simple AEB test scenarios, and have limited applicability to complex traffic scenarios; traffic flow simulation methods rely too heavily on alternative safety indicators (SSMs) calculated based on physical parameters to assess risk, such as distance-to-collision (TTC) events. This indirect assessment method has limitations in accuracy, and the indicator does not have a linear relationship with the degree of collision; research based on driving simulators often lacks collision events and focuses too much on the interaction process in human-vehicle co-driving, thus ignoring potential accident consequences. In summary, current methods for assessing the safety of autonomous vehicles throughout their operation suffer from inaccuracies. There is an urgent need for a new technical solution that can effectively collect data on autonomous vehicles in hazardous scenarios, including driving behavior, collision conditions, and occupant injury risks. This solution should propose a unified real-time safety benefit assessment model that integrates driving risk, collision risk, and occupant injury risk, thereby objectively and reasonably assessing the safety of future road traffic scenarios.

[0052] This application provides a method for assessing the safety of autonomous vehicles throughout their entire driving process, relating to the field of road traffic safety. By integrating driving risks, collision risks, and occupant injury risks in hazardous scenarios, it innovatively develops a model for real-time evaluation of the safety benefits of autonomous vehicles. This model extends the safety assessment of autonomous vehicles in normal driving scenarios to hazardous scenarios (i.e., dangerous conditions during driving), enabling the inclusion of collision risks and occupant injury risks in hazardous scenarios into the quantitative scope of overall road traffic safety. This allows for a comprehensive consideration of driving risks, collision risks, and occupant injury risks to assess the safety of autonomous vehicles. The following embodiments will specifically illustrate the method for assessing the safety of autonomous vehicles throughout their entire driving process as described in this application.

[0053] The safety assessment method for the entire driving process of an autonomous vehicle provided in this application embodiment can be applied to, for example... Figure 1 The computer device shown can be an autonomous driving system for an autonomous vehicle. Its internal structure diagram can be as follows: Figure 1As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for safety assessment throughout the entire driving process of an autonomous vehicle. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0054] Those skilled in the art will understand that Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0055] In some embodiments, such as Figure 2 As shown, a method for safety assessment of the entire driving process of an autonomous vehicle is provided, and this method is applied to... Figure 1 Taking a computer device as an example, the explanation includes the following steps:

[0056] S201, acquire the first driving data of the target vehicle, and acquire the second driving data of vehicles in the surrounding area of ​​the target vehicle.

[0057] The target vehicle can be an autonomous vehicle. According to the Society of Automotive Engineers (SAE), autonomous driving is classified into six levels (Level 0-5) based on the degree of automation and driver involvement required. Level 0 indicates that the vehicle is completely controlled by the driver, with the system only providing alerts or assistance functions, such as collision warnings or blind spot monitoring. Level 5 indicates that the vehicle can drive fully autonomously under any conditions without driver intervention; the vehicle may even lack a steering wheel or pedals. The target vehicle can be any of the Levels 0-5.

[0058] The first driving data refers to the driving data of the target vehicle, including at least one of the following: vehicle-level data, road-level data, occupant-level data, and restraint system-level data. Vehicle-level data includes at least one of the following: vehicle mass, dimensions (length and width), stiffness, speed, acceleration, heading angle, collision speed, collision angle, overlap ratio, and position. Road-level data includes at least one of the following: relative distance between two vehicles, relative speed, road surface friction coefficient, type of collision object (utility pole, guardrail, vehicle type, etc.), lane width, road curvature, and lane coordinate position. Occupant-level data includes at least one of the following: accelerator pedal force, brake pedal force, steering wheel angle, steering wheel torque, and occupant height, occupant weight, seat fore-aft position, occupant fore-aft angle, and lateral tilt angle. Restraint system-level data includes at least one of the following: driver's head airbag status (whether it is equipped, whether it has deployed), knee airbag status, and seatbelt force limit value. The second driving data refers to the driving data of vehicles in the surrounding area of ​​the target vehicle, which may include at least one of the following: vehicle-level data, road-level data, occupant-level data, and restraint system-level data.

[0059] In this embodiment, the target vehicle's mass, dimensions, width, stiffness, and seatbelt limiter values ​​can be pre-stored in the vehicle's electronic control unit (ECU) or database. When assessing the safety of the target vehicle, the computer equipment can utilize vehicle-level data, road-level data, occupant-level driving data, and restraint system-level data. Furthermore, the computer equipment can acquire secondary driving data from surrounding vehicles via a network.

[0060] Specifically, when acquiring vehicle-level data of a target vehicle, the computer can read data such as the target vehicle's mass, dimensions, and stiffness from the local electronic control unit (ECU) or database; obtain the target vehicle's speed through speed sensors; obtain the target vehicle's acceleration through acceleration sensors; obtain the target vehicle's heading angle through gyroscopes or inertial measurement units (IMUs); and when a collision occurs, the computer can obtain the collision speed through speed sensors, the collision angle through visual sensors, lidar, or collision sensors, and calculate the overlap rate by obtaining relevant features such as the target vehicle's volume and area through cameras.

[0061] When acquiring road surface data of the target vehicle, the computer can use lidar or millimeter-wave radar to measure the relative distance and relative speed between the vehicle and surrounding vehicles in real time by emitting electromagnetic waves and receiving reflected waves; it can obtain the road surface friction coefficient through tire force sensors, or obtain information such as road surface humidity and temperature through roadside sensors to calculate the road surface friction coefficient; when the target vehicle collides, it can obtain the type of collision object through cameras; and it can obtain lane width, road curvature, and lane coordinate position through GPS technology or other high-precision mapping technology.

[0062] When acquiring driving data from occupants in a target vehicle, the computer can obtain accelerator and brake pedal forces through pressure sensors on the accelerator and brake pedals, and steering wheel angle and torque through angle and torque sensors on the steering wheel. It can also identify the positions of key points such as the occupant's head and feet using in-vehicle cameras (such as those on the roof or above the center console), and estimate the occupant's height by combining the camera's installation position, angle, and calibration parameters. Furthermore, pressure sensors on the target vehicle's seats detect pressure distribution at different locations after the occupant sits down, and the occupant's weight is determined by analyzing this pressure distribution data. It should be noted that occupants include the driver and passengers. The computer can also obtain the seat's fore-aft position through position sensors on the seats. An angle sensor installed at the connection between the seat back and seat cushion can accurately measure the rotation angle of the seat back relative to the seat cushion. When the occupant adjusts the fore-aft tilt angle of the seat back, the tilt sensor can determine the tilt angle of the seat back by detecting the angle between the direction of gravity and the sensor axis, thus obtaining the occupant's fore-aft and lateral tilt angles.

[0063] When the computer acquires data on the restraint system of the target vehicle, it monitors the status of the head airbag and knee airbag in real time through the airbag control system of the target vehicle, including whether airbags are equipped and whether the airbags have deployed; and it reads the seat belt limiter value of the target vehicle from the local electronic control unit (ECU) or database.

[0064] S202, input the first driving data and the second driving data into the risk assessment model to conduct a risk assessment of the target vehicle and obtain the risk assessment result.

[0065] The risk assessment model can be a neural network model, a machine learning model, or a mathematical model. The risk assessment results include driving risk, collision risk, and damage risk; optionally, the results may also include the risk level or degree of each of these risks. Driving risk assessment helps to identify potential problems with the target vehicle during normal driving, such as speeding or illegal lane changes; collision risk assessment focuses on predicting the likelihood of a collision and the potential damage to the target vehicle after a collision; damage risk assessment further considers the potential injury to occupants after a collision.

[0066] In this embodiment, an initial risk assessment model can be pre-constructed using machine learning or deep learning algorithms. Then, a large amount of driving sample data is acquired (including data from normal driving, dangerous situations, near-collision scenarios, collision avoidance, and post-collision events). This driving sample data is then labeled, with annotations including the degree of driving risk, collision risk, and damage risk. The initial risk assessment model is then trained based on this data, adjusting its parameters to accurately predict risks in different driving scenarios and learn the relationship between different driving data and risks, thus obtaining the risk assessment model. After obtaining the first and second driving data based on the above steps, the computer device can input them into the risk assessment model. The risk assessment model calculates and analyzes the input data, using the learned rules and patterns to assess the driving risk, collision risk, and damage risk of the target vehicle, obtaining the risk assessment result. Optionally, after obtaining the first driving data and the second driving data based on the above steps, the computer equipment can perform cleaning, filtering, and calibration on the first driving data and the second driving data to remove noise and errors, thereby obtaining preprocessed first driving data and preprocessed second driving data. Then, the preprocessed first driving data and preprocessed second driving data are input into the risk assessment model to obtain the risk assessment result.

[0067] S203, input the risk assessment results into the safety assessment model to conduct a safety assessment of the target vehicle and obtain the safety assessment results.

[0068] The safety assessment results include the safety status and level of safety of the target vehicle. The safety assessment model can be a neural network model, a machine learning model, or a mathematical model.

[0069] In this embodiment, an initial safety assessment model can be constructed in advance using machine learning or deep learning algorithms. Then, a large amount of risk sample data is acquired (including data on the driving risk, collision risk, and damage risk of the target vehicle under normal driving conditions; data on the driving risk, collision risk, and damage risk of the target vehicle under dangerous conditions; data on the driving risk, collision risk, and damage risk of the target vehicle when a collision is imminent; data on the driving risk, collision risk, and damage risk of the target vehicle when avoiding a collision; and data on the driving risk, collision risk, and damage risk of the target vehicle after a collision). The risk sample data is then labeled, with the labeling information including the vehicle's safety status and safety level. The initial safety assessment model is then trained based on the risk sample data, adjusting the model's parameters to enable it to accurately predict risks in different driving scenarios and learn the relationship between different driving data and safety status and safety level, thereby obtaining the safety assessment model. After obtaining the risk assessment results based on the above steps, the computer device can input the risk assessment results into the safety assessment model. The safety assessment model performs calculations and analyses based on the input data, using the rules and patterns it has learned to evaluate the safety status and safety level of the target vehicle, thus obtaining the safety assessment result. It should be noted that the methods described in the embodiments of this application can be applied to autonomous vehicles during testing to achieve a safety assessment of the entire driving process of autonomous vehicles during testing. Subsequently, the autonomous vehicles can be operated and managed based on the safety assessment results. Optionally, the methods described in the embodiments of this application can also be applied to autonomous vehicles in real-world scenarios to achieve a safety assessment of the entire driving process of autonomous vehicles in real-world scenarios. Subsequently, traffic management departments can use the safety assessment results to conduct safety monitoring and management of traffic. Optionally, the methods described in the embodiments of this application can also be applied to other scenarios, the specific scenarios of which can be determined according to actual needs.

[0070] The method for assessing the safety of autonomous vehicles throughout their driving process provided in this application involves acquiring first driving data of the target vehicle and second driving data of vehicles in the surrounding area. This first and second driving data are then input into a risk assessment model to assess the risk of the target vehicle, yielding a risk assessment result. Finally, the risk assessment result is input into a safety assessment model to assess the safety of the target vehicle, yielding a safety assessment result. The risk assessment result includes driving risk, collision risk, and injury risk. This method utilizes the risk assessment model and the safety assessment model, combining driver behavior data and vehicle dynamics data from both normal and hazardous traffic scenarios. This extends safety assessment from hazardous scenarios forward to the normal driving process and backward to collisions and occupant injuries. It also extends safety benefit assessment from traditional driving risk quantification to collision risk quantification and occupant injury risk assessment, achieving a more comprehensive and in-depth assessment of the safety of autonomous vehicles, thereby improving the accuracy of the assessment.

[0071] In some embodiments, the risk assessment model includes a driving risk assessment sub-model, a collision risk assessment sub-model, and a damage risk assessment sub-model. Based on this, a specific implementation method for risk assessment of a target vehicle is provided, such as... Figure 3 As shown, the phrase "inputting the first driving data and the second driving data into the risk assessment model to conduct a risk assessment of the target vehicle and obtain the risk assessment result" in S202 above includes:

[0072] S301, input the first driving data into the driving risk assessment sub-model for evaluation, and obtain the driving risk.

[0073] The risk assessment model includes sub-models for driving risk assessment, collision risk assessment, and damage risk assessment. It considers the overall stability of vehicle operation and quantifies potential abnormal states during driving. Driving risks mainly include deviations from the normal driving trajectory, loss of stability, or approaching critical states.

[0074] In this embodiment, an initial driving risk assessment sub-model can be constructed in advance using machine learning or deep learning algorithms. Then, a large amount of driving sample data is acquired (including data from normal driving, dangerous situations, near-collision scenarios, collision avoidance, and post-collision events). This driving sample data is then labeled with information including driving risk. The initial driving risk assessment sub-model is then trained based on this data, adjusting its parameters to accurately predict driving risks in different driving scenarios and learning the relationship between different driving data and driving risks, thus obtaining the driving risk assessment sub-model. After obtaining the initial driving data, the computer device can input it into the driving risk assessment sub-model. The sub-model calculates and analyzes the input data, using the learned rules and patterns to assess the driving risk of the target vehicle and obtain the driving risk.

[0075] Optional, such as Figure 4 As shown, the aforementioned driving risk assessment sub-model includes a weight analysis sub-model and a first calculation sub-model. The phrase "inputting the first driving data into the driving risk assessment sub-model for evaluation to obtain the driving risk" in S301 includes:

[0076] S3011, input the first driving data into the weight analysis sub-model to obtain the weight of the target factor.

[0077] The driving risk assessment sub-model includes a weighted analysis sub-model and a first calculation sub-model. Target factors include vehicle speed, acceleration, lane departure distance, and environmental factors. The weighted analysis sub-model is obtained by least squares fitting.

[0078] In this embodiment of the application, a weight analysis sub-model can be pre-constructed based on the least squares method. After obtaining the first driving data, the computer device can input information such as the speed, acceleration, whether an accident has occurred, and whether it is in a dangerous state of the target vehicle in the first driving data into the weight analysis sub-model to obtain the weight of the target factor.

[0079] S3012, input the weights of the target factors into the first calculation sub-model to obtain the driving risk.

[0080] In this embodiment of the application, a first calculation sub-model can be pre-constructed based on factors such as vehicle speed, acceleration, lane departure distance, environmental factors, and the weights corresponding to each factor, as detailed below:

[0081]

[0082]

[0083] in, Indicates driving risk; The standard deviation of vehicle speed fluctuation (i.e., vehicle speed factor); This represents lateral acceleration (i.e., acceleration factor). The lateral deviation distance from the lane (i.e., the lane deviation factor) is determined by the vehicle's position coordinates and the lane line coordinates. Indicates environmental complexity (i.e., environmental factors); This represents the weighting coefficient for each objective factor; Indicates traffic flow density; Indicates the curvature of the road; Indicates weather conditions (W=1 indicates sunny, W=1.5 indicates light rain or light snow, W=2 indicates heavy rain, heavy snow, or dense fog).

[0084] S302, input the first driving data and the second driving data into the collision risk assessment sub-model for evaluation to obtain the collision risk.

[0085] Collision risk is determined by the interaction between the vehicle and other vehicles in the environment during the vehicle's operation.

[0086] In this embodiment, an initial collision risk assessment sub-model can be constructed in advance using machine learning or deep learning algorithms. Then, a large amount of driving sample data (including data on near-collision moments, collision avoidance data, and post-collision data) is acquired. This driving sample data is then labeled with collision risk information. The initial collision risk assessment sub-model is then trained based on this data, adjusting its parameters to accurately predict collision risks in different driving scenarios. The model learns the relationship between different driving data and driving risks, thus obtaining the collision risk assessment sub-model. After obtaining the first and second driving data, the computer device can input them into the collision risk assessment sub-model. The sub-model calculates and analyzes the input data, using the learned rules and patterns to assess the collision risk of the target vehicle and obtain the collision risk.

[0087] Optional, such as Figure 5 As shown, the aforementioned collision risk assessment sub-model includes an analysis sub-model and a second calculation sub-model. The phrase "inputting the first driving data and the second driving data into the collision risk assessment sub-model for assessment to obtain the collision risk" in S302 includes:

[0088] S3021, input the first driving data and the second driving data into the analysis sub-model to obtain the longitudinal collision risk and the lateral collision risk.

[0089] The collision risk assessment sub-model includes an analysis sub-model and a second calculation sub-model.

[0090] In this embodiment, an analysis sub-model can be pre-constructed, specifically as follows: Based on the numerous environmental vehicles surrounding the target vehicle, vehicles within a preset distance (e.g., 100 meters) are defined as "dangerous environmental vehicles," otherwise they are defined as "safe environmental vehicles." For each dangerous environmental vehicle, the longitudinal collision risk and lateral collision risk between it and the target vehicle are determined. Specifically, the longitudinal collision risk, using TTC as an example, represents the remaining time before a collision occurs when the target vehicle and the dangerous environmental vehicle maintain their current driving state (speed, heading angle). The lateral collision risk is determined by the lateral distance between the target vehicle and the dangerous environmental vehicle (…). The longitudinal collision risk is represented by (). The specific mathematical expression for longitudinal collision risk is as follows:

[0091]

[0092] in, This represents the relative distance between the vehicle and another vehicle in the environment at time t. Let t represent the relative speed between the vehicle and the vehicle in the environment in the direction of the vehicle's travel.

[0093] After obtaining the first and second driving data, the computer equipment can input the first and second driving data into the analysis sub-model. The first and second driving data can be used to objectively quantify the safety at a certain moment and obtain the longitudinal collision risk and lateral collision risk of the target vehicle.

[0094] S3021, input the longitudinal collision risk and the lateral collision risk into the second calculation sub-model to obtain the collision risk.

[0095] In this embodiment, a second calculation sub-model can be pre-constructed based on longitudinal collision risk and lateral collision risk, as follows: When the time is less than a pre-set time threshold (e.g., TTC < 2 seconds), the longitudinal collision risk is taken inversely and integrated over the duration less than the threshold to obtain the collision risk. ,as follows:

[0096]

[0097]

[0098] Where RT represents the road type; This represents the weight coefficient of each factor; t represents the duration for which TTC is less than the pre-set threshold. Indicates collision risk. The parameters in the model can be determined by fitting methods, such as the least squares method, based on real accident statistics of vehicles with different levels of autonomous driving (collision rate as model output, detailed collision conditions, driving mileage and other parameters as model input).

[0099] After obtaining the longitudinal and lateral collision risks, the computer equipment calculates the target vehicle's collision risk based on the longitudinal and lateral collision risks if the longitudinal collision risk is less than 2 seconds. If the longitudinal collision risk is not less than 2 seconds, the target vehicle is determined to have no collision risk, i.e., the target vehicle's collision risk is zero.

[0100] S303, input the first driving data and the second driving data into the damage risk assessment sub-model for evaluation, and obtain the damage risk.

[0101] Among them, damage risk represents the severity of occupant injury and is an important basis for safety benefit assessment. The core protection object of autonomous vehicles is the occupants.

[0102] In this embodiment of the application, a damage risk assessment sub-model can be obtained in advance. After obtaining the first driving data and the second driving data, the computer device can input the first driving data and the second driving data into the damage risk assessment sub-model for assessment to obtain the damage risk.

[0103] Specifically, such as Figure 6 As shown, the method for obtaining the above-mentioned damage risk assessment sub-model includes:

[0104] S3031, acquire collision speed sample data and other sample data of the vehicle under collision conditions.

[0105] Other sample data include at least one of the following: initial collision condition sample data, occupant physiological parameter sample data, and occupant posture sample data.

[0106] In this embodiment of the application, the computer can obtain collision speed sample data and other sample data of vehicles under collision conditions by web crawling or by using publicly available traffic websites.

[0107] S3032, perform high-dimensional processing on other sample data to obtain processed other sample data, and train the initial damage risk assessment sub-model based on the processed other sample data to obtain the damage risk assessment sub-model.

[0108] The initial damage risk assessment sub-model can be either a neural network model or a machine learning model.

[0109] In this embodiment of the application, after the computer device obtains the collision speed sample data and other sample data of the vehicle under the collision condition based on the above steps, since the collision speed is a key quantitative indicator under the collision condition, in order to ensure the accuracy of the damage risk assessment sub-model, no high-dimensional processing is performed on the collision speed sample data, and the original values ​​are retained. Only the other sample data is processed to obtain the processed other sample data. Then, the initial damage risk assessment sub-model is trained based on the processed other sample data to obtain the damage risk assessment sub-model.

[0110] For example, in hazardous scenarios, computer equipment can determine the approximate future collision conditions based on the relative position and speed of vehicles at the current moment. It is worth noting that even in the event of a collision, vehicles at different levels of autonomous driving exhibit significant differences in occupant protection due to variations in their avoidance strategies, safety device optimizations, and driver habits. Key variables significantly influencing occupant injury can be summarized into three aspects: people, vehicle, and road. Specifically, these include: people—the physiological characteristics of vehicle occupants (age, gender, height, weight, etc.) and seating posture (seat orientation, backrest angle, body posture); vehicle—vehicle physical parameters (mass, dimensions, stiffness, etc.) and restraint system information (seat belts, airbags, etc.); and road—obstacle types (cars, pedestrians, buildings), collision speed, and collision location. Therefore, the severity of occupant injury is rapidly and dynamically influenced by the interaction and coupling of these multiple factors. The initial collision condition input and occupant injury output exhibit a complex, multi-layered mapping relationship, making it difficult to accurately establish a mathematical model.

[0111] In summary, a neural network-based occupant injury risk prediction model can be used to quantify the severity of occupant injuries in each potential collision or collision accident. Computer equipment can acquire collision speed sample data and other sample data under collision conditions. These other sample data include at least one of the following: initial collision condition sample data, occupant physiological parameter sample data, and occupant posture sample data. Then, a neural network model based on a multilayer perceptron (MLP) can be used to directly predict the severity of occupant injuries during the collision based on the specific conditions at the moment of collision, occupant physiological parameters, constraint system parameters, and occupant posture data. Taking the severity of head injuries as an example, the structure of the neural network model is as follows: Figure 7 As shown, the severity of other parts is determined using the same algorithm:

[0112] The model consists of a data preprocessing module, a neural network processing module, and a result postprocessing module.

[0113] Data preprocessing module (input layer): Since the initial collision conditions (collision speed, collision angle, vehicle offset, ground friction coefficient), occupant physiological parameters (gender, height, weight), restraint system (seat belts, head airbags, knee airbags), and occupant posture (sitting position, attitude, distance from steering wheel) are all scalars, the remaining scalar information, except for collision speed, is mapped to a high-dimensional space that is conducive to neural network processing through the embedding layer, thereby enriching the features of the input data. Because there is a significant quantitative relationship between collision speed and occupant injury (the higher the collision speed, the more severe the occupant injury tends to be), the model chooses to use collision speed as a direct input without processing through the embedding layer.

[0114]

[0115]

[0116] in, It is scalar information other than collision velocity. It is the high-dimensional feature vector after embedding. It is the collision speed. It is a feature embedding function. It is a feature concatenation function. It is the input feature matrix.

[0117] Neural network processing module (hidden layer): The model consists of two fully connected layers and two normalized layers. The activation function can be ReLU or tanh, etc.

[0118] a. Fully connected layer 1:

[0119]

[0120] b. Normalization layer 1:

[0121]

[0122] c. Fully Connected Layer 2:

[0123]

[0124] d. Normalized layer 2:

[0125]

[0126] in, and These are the weight matrix and bias vector of the first fully connected layer. It is the output of the first layer. and They are The mean and variance; and These are the weight matrix and bias vector of the second fully connected layer. It is the output of the second layer.

[0127] The post-processing module (output layer) consists of a linear layer that maps the information processed by the neural network from a high-dimensional space to a low-dimensional space. After discretization, it finally outputs the occupant's head injury criterion (HIC).

[0128]

[0129]

[0130] in, and These are the weight matrix and bias vector of the linear layer. It is a low-dimensional feature vector. It is a discretization processing function.

[0131] Model Training and Optimization: The database used for training the neural network model can be generated through joint simulation using finite element method (FEM) software and multibody dynamics simulation software (such as MADYMO). First, the parameter information for the specific collision condition is determined. A high-precision vehicle model is constructed using FEM software (such as LS-DYNA or ANSYS), and intermediate parameters such as the collision waveform are calculated and generated. The collision waveform is then input into the multibody dynamics model (MADYMO), and occupant and constraint system parameters are configured to calculate and output the damage to various parts of the occupants' bodies. This results in the construction of a database containing specific collision conditions and the severity of occupant injuries.

[0132] The damage function can be either the mean squared error damage function or the cross-entropy loss function. The model optimizer can use stochastic gradient descent (SGD, Adam, etc.). The mean squared error loss function can be expressed as:

[0133]

[0134] in, It is the actual occupant injury value (e.g., head injury risk factor). This is the predicted occupant injury value.

[0135] The aforementioned neural network can quantify the severity of injuries to different parts of the body. This can be represented by the Abbreviated Injury Scale (AIS), and the risk of occupant injury in hazardous scenarios can be obtained through weighted summation.

[0136]

[0137] in, This represents the weight of different parts of the body, for example, the weight that can be taken. , , .

[0138] The method described in this application not only performs data collection, model building, and preliminary verification in a driving simulator, but also designs a migration strategy from the simulated environment to real road traffic scenarios. The model design inputs consider the feasibility of real-vehicle deployment, including easily accessible multi-dimensional vehicle dynamics data during daily vehicle operation, as well as operational characteristic data during hazardous situations and collisions. All of these data can be used as inputs to the safety benefit assessment model. During model development and optimization, a natural driving database and a real accident database are introduced to improve the model's accuracy and reliability. By analyzing the massive amount of driving behavior data in the natural driving database, specific driving behaviors and thresholds of actual drivers in different road traffic situations can be extracted, thereby limiting the thresholds of model weight parameters in specific road traffic scenarios and providing the model with parameter settings closer to real driving behavior. Simultaneously, statistical data, accident characteristics, and final economic losses (including property damage and medical insurance losses due to personal injury) from the real accident database are combined. The overall economic loss is used as an objective quantification of the unified safety benefit in the model, thereby adjusting the weight values ​​of driving risk, collision risk, and occupant injury risk in the model, making the weight parameters more consistent with industry statistical data and reports for autonomous vehicles. This fine-tuning process based on real-world data not only enhances the model's accuracy in assessing the unified safety benefits of vehicles at different levels of autonomous driving based on driving risk, collision risk, and occupant injury risk, but also verifies the model's performance under imperfect inputs by using outliers (noise, missing data) in real-world data, ensuring the model's robustness. After successful model validation, it is deployed to large-scale fleets and traffic flows for real-world testing. During this process, driving behavior and hazardous condition data are collected in real time through onboard systems or roadside connected devices, monitoring the unified safety benefits output by the model. This ensures the feasibility and universality of the technical solution in real-world road scenarios, enabling the assessment of safety benefits for vehicles at different levels of autonomous driving to operate in real-time and long-term in the real world, thereby improving road traffic safety. It also enables the collection of driving behavior data and collision condition data serving occupant injury risk prediction in hazardous scenarios. By employing a high-fidelity driving simulator and a multi-dimensional data acquisition system, comprehensive data on vehicles, scenarios, and drivers under different conditions is collected. A neural network model is then used to rapidly quantify the severity of occupant injuries in potential collisions under hazardous scenarios, ensuring the completeness of data in these scenarios and the reliability of input data for the safety benefit assessment model. A real-time safety benefit assessment model for unified driving risk, collision risk, and occupant injury risk under hazardous scenarios is constructed. This invention develops a unified safety benefit assessment model applicable to vehicles of different levels of autonomous driving, covering normal driving scenarios, hazardous scenarios, and collision accident scenarios.It proposes a technical path from data verification based on driving simulators, to fine-tuning of model parameters based on natural driving data and real accident data, and finally to large-scale real vehicle application. This ensures that the model can accurately and objectively quantify the safety benefits of different levels of autonomous driving vehicles in various road traffic scenarios, providing an objective safety basis for the development of autonomous vehicles.

[0139] In some embodiments, the method for assessing the safety of an autonomous vehicle throughout its driving process includes:

[0140] S401, Obtain the safety assessment scale.

[0141] The safety assessment scale includes multiple driving factors, each driving factor includes multiple driving safety questions, and each driving safety question has a corresponding score.

[0142] In this embodiment, a large number of safety assessment scales can be obtained by having volunteers participate in simulated experiments, or by selecting safety-related content from the commonly used autonomous vehicle acceptance scale in existing research, and supplementing the scale with the "perceived safety" section and the "experimental validity" section to determine the validity of the experiment and scale, based on the characteristics of dangerous scenarios in the experiment, to form a large number of self-safety assessment scales. The safety assessment scale contains 43 items. Participants are required to read each item and score it according to the degree of consistency between the item and their feelings, thoughts and behaviors when driving the corresponding autonomous driving level vehicle. The scoring uses a 5-point scale, ranging from "absolutely disagree (1)" to "absolutely agree (5)". Six items are used to determine whether the scale is valid (Q38-Q40: verifying the consistency between behavior in the experiment and real driving; Q41-Q43: verifying the authenticity and closeness of the experimental scenario). The 36 items are classified into 6 factors, each of which represents a specific characteristic description of the subjective quantification of autonomous vehicle safety. The subjective safety assessment of different autonomous driving levels of vehicles can be quantified in a coordinated manner based on the scores of different factors. The specific safety assessment scale is shown in the table below:

[0143]

[0144] S402, based on the safety assessment scale and driving risk parameters, collision risk parameters and damage risk parameters, construct a safety assessment model.

[0145] In this embodiment of the application, after the computer device obtains the safety assessment scale based on the above steps, it can construct a safety assessment model based on the safety assessment scale, driving risk parameters, collision risk parameters, and damage risk parameters.

[0146] Specifically, the phrase "constructing a safety assessment model based on the safety assessment scale and driving risk parameters, collision risk parameters, and damage risk parameters" in S402 above includes:

[0147] S4031, Construct a driving factor load matrix based on the safety assessment scale.

[0148] The driving factor loading matrix represents the correlation (positive correlation, negative correlation, or no correlation) between each item.

[0149] S4032, determine the load value corresponding to each driving factor based on the driving factor load matrix.

[0150] S4033, based on the load values ​​corresponding to each driving factor and the scores corresponding to each driving safety question, determine the weights of the driving risk parameters, collision risk parameters, and damage risk parameters.

[0151] In this embodiment, after obtaining the safety assessment scale (hereinafter referred to as the scale), the computer device first evaluates the effectiveness of the experiment for the volunteer by scoring the "experiment validity" section (questions 38-43) in the safety assessment scale. If the score is higher than a preset value (e.g., 80 points), it indicates that the safety assessment scale is valid, and the scale can be used for further analysis. If it is invalid, the scale is deleted, and other safety assessment scales are re-evaluated for validity. Secondly, the volunteer's behavior in the experiment can be verified to be consistent with the dimensions measured by the scale through other parts (overall trust, driving / riding emotions, perceived safety, perceived usefulness, and perceived ease of use). Specifically, this can be achieved by obtaining driving behavior data of the volunteer during the driving process (e.g., accelerator pedal force). ), brake pedal force ( ), Steering wheel angle ( Steering wheel torque ( Correlation analysis was performed between this data and the overall trust level, driving / riding emotions, and perceived ease of use on the scale (e.g., verifying the correlation between the occurrence of behaviors such as emergency braking and sharp steering and driving emotions). For example, the brake pedal force recorded in the experiment under dangerous scenarios ( This can be directly compared with the scores of the worry or anxiety items in the driving / riding emotion section of the scale. For example, if a volunteer always exerts braking force while driving, it indicates that the emotion during driving / riding is worry or anxiety. In this case, it can be checked whether there is a positive correlation between the exertion of braking force and worry or anxiety. Finally, after the scale is statistically analyzed, the item scores are validated to ensure that the measured variables conform to the expected psychological or statistical logic. For example, a relaxed driving mood and trust in autonomous driving technology should be positively correlated (for example, the linear relationship of variables can be verified by calculating Pearson correlation).

[0152] To quantify the weighting of driving risk, collision risk, and damage risk on the unified safety benefit, the following method is used:

[0153] ① The scores for each item in the scale are normalized to a mean of 0 and a variance of 1 to eliminate differences in the scoring scales of different items, making the scores of different subjects comparable.

[0154]

[0155] in, This indicates the subject's score on that question. This represents the mean of each question. This represents the standard deviation for each item.

[0156] ② Factor analysis: Calculate the factor loadings for each question.

[0157] a. First, calculate the correlation matrix R (i.e., the driving factor loading matrix) of the scale data, representing the correlation (positive correlation, negative correlation, or no correlation) between each item:

[0158]

[0159] in:

[0160]

[0161]

[0162]

[0163] For each pair of items (i, j), calculate the Pearson correlation coefficient:

[0164]

[0165] in, This is the standardized score for question i; is the standardized score of item j; n is the number of participants or the number of safety assessment scales.

[0166] b. Perform Principal Component Analysis (PCA): Calculate eigenvalues, select factors with eigenvalues ​​> 1 according to the Kaiser criterion, obtain the loading matrix F, and then perform Varimax rotation to obtain the final driving factor loading matrix. (After rotation, the factor loadings are clearer), and from Extract the loading values ​​of the corresponding factors ,in, include , , … , , , This can be represented as m×n, where m represents the question and n represents the factor (i.e., the first column "Factors" in the safety assessment scale), the driving factor loading matrix. for:

[0167]

[0168] ③ Calculate the weight of each risk (driving, collision, damage) on the overall safety benefit.

[0169] Based on the scale content, select questions related to driving, collision, and injury risks from the items, and combine these with the scale scores. Quantify the weights of driving, collision, and damage risks:

[0170] The following standards can be used as a reference for classifying questions:

[0171] Driving risks: 1-2, 4-7, 15-16, 19, 21-22, 24, 29;

[0172] Collision risk: 5-7, 14, 18, 20, 23, 27-33;

[0173] Damage risk: 3-7, 16-18, 25-28, 30;

[0174]

[0175]

[0176]

[0177] in, Indicates the weight of driving risk parameters; Indicates the weights of the collision risk parameters; This represents the weight of the damage risk parameter.

[0178] S5034, a safety assessment model is constructed based on the weights of driving risk parameters, collision risk parameters, damage risk parameters, driving risk parameters, collision risk parameters, and damage risk parameters.

[0179] In this embodiment, after the computer device obtains the weights of the driving risk parameter, the collision risk parameter, and the damage risk parameter based on the above steps, it can construct a safety assessment model based on these weights. The specific safety assessment model is as follows:

[0180]

[0181] in, Indicates the safety assessment value; Indicates driving risk; Indicates the risk of collision; Indicates the risk of damage; Indicates driving risk parameters; Indicates collision risk parameters; Indicates damage risk parameters; Indicates the total mileage; This indicates the total travel time within the preset area.

[0182] The method described in this application embodiment can effectively collect data on the entire process of autonomous vehicles in dangerous scenarios, including driving behavior, collision conditions, and occupant injury risks. It proposes a unified real-time safety benefit assessment model that integrates driving risk, collision risk, and occupant injury risk, thereby objectively and reasonably assessing the safety of future road traffic scenarios.

[0183] In summary, based on all the above embodiments, a method for safety assessment of the entire driving process of an autonomous vehicle is also provided, the method comprising:

[0184] S601, Obtain the safety assessment scale, construct a driving factor load matrix based on the scale, determine the load value for each driving factor, and determine the weights of the driving risk parameter, collision risk parameter, and damage risk parameter based on the load values ​​and scores for each driving safety question. Construct a safety assessment model based on these weights. The safety assessment scale includes multiple driving factors, each driving factor includes multiple driving safety questions, and each driving safety question has a corresponding score.

[0185] S602, acquire collision speed sample data and other sample data of the vehicle under collision conditions, perform high-dimensional processing on the other sample data to obtain processed other sample data, and train an initial damage risk assessment sub-model based on the processed other sample data to obtain the damage risk assessment sub-model. The other sample data includes at least one of the following: initial collision condition sample data, occupant physiological parameter sample data, and occupant posture sample data.

[0186] S603, acquire the first driving data of the target vehicle, and acquire the second driving data of vehicles in the surrounding area of ​​the target vehicle.

[0187] S604: Input the first driving data into the weight analysis sub-model to obtain the weights of the target factors. Then, input the weights of the target factors into the first calculation sub-model to obtain the driving risk. The target factors include vehicle speed, acceleration, lane departure distance, and environmental factors.

[0188] S605, input the first driving data and the second driving data into the analysis sub-model to obtain the longitudinal collision risk and the lateral collision risk, and input the longitudinal collision risk and the lateral collision risk into the second calculation sub-model to obtain the collision risk.

[0189] S606, input the first driving data and the second driving data into the damage risk assessment sub-model for evaluation, and obtain the damage risk.

[0190] S607 inputs driving risk, collision risk, and damage risk into the safety assessment model to conduct a safety assessment of the target vehicle and obtain the safety assessment results.

[0191] In this embodiment, vehicle driving risk refers to the fact that all components of the traffic system have the potential to generate driving risks. The risks posed to vehicles on the road are objective and independent of human will. During vehicle operation, safety is affected by numerous factors, and these factors (such as driver behavior characteristics, vehicle driving status, and road traffic environment) are dynamically changing. Therefore, the probability, degree, and category of driving risk are also dynamically changing and uncertain. The driving risk of autonomous vehicles reflects their safety and reliability in complex and variable situations in real traffic environments. Vehicle collision risk and accident conditions: Vehicle collision risk refers to the possibility and severity of potential consequences of a physical collision with other vehicles, pedestrians, objects, or obstacles during dynamic driving, due to the influence of the external environment, vehicle performance, or driving behavior. Collision risk assessment is an important indicator for measuring vehicle safety performance; it can help predict and reduce accidents and improve driving safety. Accident conditions describe the environmental conditions and dynamic response of the vehicle during a collision accident, as well as related physical parameters. Specifically, parameters include those at the vehicle level (collision speed, angle, location, overlap rate), road level (road surface friction coefficient, nature of the colliding object), occupant level (height, weight), and restraint system level (seatbelt usage, seatbelt force limiter, airbag deployment). These specific conditions can be used to analyze the causes of collision accidents and changes in vehicle dynamics, thereby further quantifying the severity of collision accidents and the potential injury risk to occupants, providing important basis for accident reconstruction and safety assessment. Occupant injury risk: Taking traffic participants as the protected objects, injury refers to the deformation of human anatomical structures under external loads exceeding their failure limits, leading to tissue damage or structural failure. The injury mechanisms and failure limits of key body parts are important data references for vehicle safety assessment and traffic participant protection. At the moment of a collision, the initial kinetic energy carried by the vehicle under normal driving conditions is rapidly released and transferred to the vehicle body and occupants, causing deformation or damage, resulting in vehicle damage and occupant injury risk. The type and severity of injuries sustained by traffic participants are difficult to obtain and assess directly. They require conversion into injury indicators corresponding to specific body parts using readily measurable physical parameters such as acceleration, velocity, displacement, compression or elongation, force, and stress. Commonly used evaluation indicators include the Head Injury Criterion and the Max Deflection.

[0192] The methods described in this application lack comprehensive data covering normal driving, hazardous scenarios, and collision accidents. In safety research on autonomous vehicles, assessing their safety benefits requires multi-dimensional, end-to-end data support. However, existing data typically relies on natural driving databases lacking hazardous scenarios and real accident databases with limited data and incomplete occupant injury information. Fragmented data collection methods and single-source data with low information completeness cannot fully utilize all key safety-related information across the entire time dimension of road traffic scenarios, making it difficult to provide data support for accurately quantifying the safety benefits of autonomous vehicles. Furthermore, existing technologies lack a unified safety benefit assessment model that integrates driving, collision, and occupant injury risks. Existing evaluation standards for the safety benefits of autonomous vehicles are relatively singular, mostly quantifying only driving and collision risks, neglecting the risk of occupant injury in collision accidents under hazardous scenarios. Occupants are a key protection target for intelligent vehicles and a significant factor influencing the safety of autonomous vehicles. Therefore, this application, based on driving behavior under hazardous scenarios, comprehensively considers driving risks, collision risks, and occupant injury risks in hazardous scenarios, which helps to assess the safety benefits of autonomous vehicles in real-time and objectively.

[0193] This application presents a data acquisition system for the entire "driving-collision-damage" process in hazardous scenarios. This technology designs normal and hazardous scenarios with different traffic risk levels, develops control algorithms for vehicles with different levels of autonomous driving, and conducts driver experiments and full-process data acquisition based on a driving simulator. This invention's multi-scenario, multi-autonomous driving level and data acquisition system is an integrated solution designed to comprehensively collect and analyze driver behavior data when driving vehicles with different levels of autonomous driving in different traffic scenarios, as well as collision condition data in hazardous scenarios, using a high-fidelity driving simulator. The core of the system includes a multi-degree-of-freedom, high-fidelity driving simulator equipped with advanced dynamic feedback mechanisms and multi-dimensional data acquisition technology, and integrates control algorithms adapted to vehicles with different levels of autonomous driving. The driving simulator is designed with a focus on a realistic driving experience, supporting comprehensive dynamic simulation, including vehicle physical responses such as acceleration, braking, steering, and road conditions. The simulator also provides multi-sensory feedback close to the actual driving experience, including visual, auditory, and tactile feedback, ensuring that drivers obtain a highly realistic driving experience in a virtual environment. The system's data acquisition scope is extensive and comprehensive, covering multiple dimensions. Specifically, this includes: vehicle information collection (kinematic and dynamic data of the vehicle under normal driving, hazardous scenarios, and at the moment of collision, dynamic information of surrounding vehicles, and data captured by onboard cameras), road segment information collection (the urgency of traffic scenarios, the type of hazardous scenarios, road environmental characteristics, and ground friction coefficient, etc.), and driver information collection (collecting driving behavior, seating posture, and restraint system status through motion capture and depth imaging technology, combined with age, driving experience, and other identity information for research).

[0194] This application develops a real-time safety benefit assessment model for autonomous vehicles. This invention innovatively combines driving risk and collision risk during normal driving in hazardous scenarios, and quantifies the risk of occupant injury in potential collisions through an occupant injury risk prediction model, for real-time, comprehensive assessment of the safety benefits of autonomous vehicles in hazardous scenarios. This application considers model verification and application in real-world road traffic environments. This invention utilizes normal driving data based on a driving simulator, driving behavior data in hazardous scenarios, collision condition data, and an occupant injury risk prediction model to establish a real-time safety benefit assessment model for autonomous vehicles in hazardous scenarios, and verifies the effectiveness of the model. Finally, it proposes a transfer strategy from the simulated environment to real-world road traffic scenarios. Based on a high-fidelity driving simulator, this technical solution is applicable to vehicles with different levels of autonomous driving, simulating normal traffic scenarios and various hazardous traffic scenarios, and collecting driver behavior characteristics when driving vehicles with different levels of autonomous driving. This invention utilizes an occupant injury risk prediction model, combined with driver behavior data and vehicle dynamics data from both normal and hazardous traffic scenarios, to develop a novel safety benefit assessment model for autonomous vehicles in hazardous scenarios. This model extends safety assessment forward from hazardous scenarios to the normal driving process and backward to collisions and occupant injuries. It also extends safety benefit assessment from traditional driving risk quantification to collision risk quantification and occupant injury risk assessment, exploring the relationship between the safety benefits of autonomous vehicles and driving behavior in hazardous scenarios. This invention mainly comprises three parts: a full-process data acquisition system for hazardous scenarios, a unified safety benefit assessment model for autonomous vehicles, and model verification and application considering real-world road traffic environments.

[0195] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0196] Based on the same inventive concept, this application also provides an autonomous vehicle driving process safety assessment device for implementing the above-mentioned autonomous vehicle driving process safety assessment method. The solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations of one or more embodiments of the autonomous vehicle driving process safety assessment device provided below can be found in the limitations of the autonomous vehicle driving process safety assessment method above, and will not be repeated here.

[0197] In some embodiments, such as Figure 8 As shown, a safety assessment device for the entire driving process of an autonomous vehicle is provided, comprising:

[0198] The acquisition module is used to acquire first driving data of the target vehicle and second driving data of vehicles in the surrounding area of ​​the target vehicle.

[0199] The risk assessment module is used to input first and second driving data into the risk assessment model to conduct a risk assessment of the target vehicle and obtain the risk assessment results; the risk assessment results include driving risk, collision risk and damage risk.

[0200] The safety assessment module is used to input the risk assessment results into the safety assessment model to conduct a safety assessment of the target vehicle and obtain the safety assessment results.

[0201] In some embodiments, the risk assessment module described above includes:

[0202] The first assessment unit is used to input the first driving data into the driving risk assessment sub-model for assessment to obtain the driving risk.

[0203] The second assessment unit is used to input the first driving data and the second driving data into the collision risk assessment sub-model for assessment to obtain the collision risk.

[0204] The third assessment unit is used to input the first driving data and the second driving data into the damage risk assessment sub-model for assessment to obtain the damage risk.

[0205] In some embodiments, the first evaluation unit is specifically used to input the first driving data into the weight analysis sub-model to obtain the weights of the target factors; the target factors include vehicle speed factors, acceleration factors, lane departure distance factors, and environmental factors; and input the weights of the target factors into the first calculation sub-model to obtain the driving risk.

[0206] In some embodiments, the second evaluation unit is specifically used to input the first driving data and the second driving data into the analysis sub-model to obtain the longitudinal collision risk and the lateral collision risk; and to input the longitudinal collision risk and the lateral collision risk into the second calculation sub-model to obtain the collision risk.

[0207] In some embodiments, the risk assessment module described above is further used to acquire collision speed sample data and other sample data of the vehicle under collision conditions; the other sample data includes at least one of initial collision condition sample data, occupant physiological parameter sample data and occupant posture sample data; the other sample data is subjected to high-dimensional processing to obtain processed other sample data, and the initial damage risk assessment sub-model is trained based on the processed other sample data to obtain the damage risk assessment sub-model.

[0208] In some embodiments, the above-mentioned autonomous vehicle driving safety assessment device is further used to obtain a safety assessment scale; the safety assessment scale includes multiple driving factors, each driving factor includes multiple driving safety questions, and the scores corresponding to each driving safety question; and a safety assessment model is constructed based on the safety assessment scale and driving risk parameters, collision risk parameters, and damage risk parameters.

[0209] In some embodiments, the above-mentioned autonomous vehicle driving safety assessment device is further configured to construct a driving factor load matrix based on a safety assessment scale; determine the load value corresponding to each driving factor based on the driving factor load matrix; determine the weights of driving risk parameters, collision risk parameters, and damage risk parameters based on the load values ​​corresponding to each driving factor and the scores corresponding to each driving safety question; and construct a safety assessment model based on the weights of driving risk parameters, collision risk parameters, and damage risk parameters.

[0210] Each module in the aforementioned safety assessment device for the entire driving process of autonomous vehicles can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0211] In some embodiments, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the safety assessment method for the entire driving process of an autonomous vehicle as described in any of the above embodiments.

[0212] In some embodiments, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the safety assessment method for the entire driving process of an autonomous vehicle as described in any of the above embodiments.

[0213] In some embodiments, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the safety assessment method for the entire driving process of an autonomous vehicle as described in any of the above embodiments.

[0214] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile 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 many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0215] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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 specification.

[0216] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for safety assessment of the entire driving process of an autonomous vehicle, characterized in that, The method includes: Acquire first driving data of the target vehicle, and acquire second driving data of vehicles in the surrounding area of ​​the target vehicle; The first driving data and the second driving data are input into a risk assessment model to conduct an objective risk assessment of the target vehicle throughout its driving process, resulting in a risk assessment result. The risk assessment result includes driving risk, collision risk, and damage risk. The driving risk is used to identify potential problems that may occur to the target vehicle during normal driving. The collision risk is used to predict the likelihood of a collision occurring to the target vehicle during driving and the extent of damage to the target vehicle after a collision. The damage risk is used to assess the extent of injury to occupants after a collision. The driving risk is determined by vehicle speed fluctuations, lateral acceleration, lane lateral deviation distance, and environmental complexity. A safety assessment scale is obtained, and a safety assessment model is constructed based on the safety assessment scale and driving risk parameters, collision risk parameters, and damage risk parameters. The safety assessment scale includes multiple driving safety items, each representing the driver's subjective safety assessment of vehicles with different levels of autonomous driving. The safety assessment scale is used to determine the weights of the driving risk parameters, the collision risk parameters, and the damage risk parameters. The risk assessment results are input into the safety assessment model to conduct a safety assessment of the target vehicle, and the safety assessment results are obtained.

2. The method according to claim 1, characterized in that, The risk assessment model includes a driving risk assessment sub-model, a collision risk assessment sub-model, and a damage risk assessment sub-model. The step of inputting the first driving data and the second driving data into the risk assessment model to perform a risk assessment on the target vehicle and obtain the risk assessment result includes: The first driving data is input into the driving risk assessment sub-model for evaluation to obtain the driving risk; The first driving data and the second driving data are input into the collision risk assessment sub-model for evaluation to obtain the collision risk. The first driving data and the second driving data are input into the damage risk assessment sub-model for evaluation to obtain the damage risk.

3. The method according to claim 2, characterized in that, The driving risk assessment sub-model includes a weight analysis sub-model and a first calculation sub-model. The step of inputting the first driving data into the driving risk assessment sub-model for evaluation to obtain the driving risk includes: The first driving data is input into the weight analysis sub-model to obtain the weights of the target factors; the target factors include vehicle speed, acceleration, lane departure distance, and environmental factors. The weights of the target factors are input into the first calculation sub-model to obtain the driving risk.

4. The method according to claim 2, characterized in that, The collision risk assessment sub-model includes an analysis sub-model and a second calculation sub-model. The first driving data and the second driving data are input into the collision risk assessment sub-model for evaluation to obtain the collision risk, including: The first driving data and the second driving data are input into the analysis sub-model to obtain the longitudinal collision risk and the lateral collision risk; The longitudinal collision risk and the lateral collision risk are input into the second calculation sub-model to obtain the collision risk.

5. The method according to claim 2, characterized in that, The method for obtaining the damage risk assessment sub-model includes: Acquire collision speed sample data and other sample data of the vehicle under collision conditions; the other sample data includes at least one of initial collision condition sample data, occupant physiological parameter sample data and occupant posture sample data. The other sample data is subjected to high-dimensional processing to obtain processed other sample data, and an initial damage risk assessment sub-model is trained based on the processed other sample data to obtain the damage risk assessment sub-model.

6. The method according to claim 1, characterized in that, The safety assessment model is constructed based on the safety assessment scale and driving risk parameters, collision risk parameters, and damage risk parameters, including: Construct a driving factor load matrix based on the aforementioned safety assessment scale; The load value corresponding to each driving factor is determined based on the driving factor load matrix; The weights of the driving risk parameter, the collision risk parameter, and the damage risk parameter are determined based on the load values ​​corresponding to each driving factor and the scores corresponding to each driving safety question. The safety assessment model is constructed based on the weights of the driving risk parameter, the collision risk parameter, the damage risk parameter, the driving risk parameter, the collision risk parameter, and the damage risk parameter.

7. A safety assessment device for the entire driving process of an autonomous vehicle, characterized in that, The apparatus for implementing the safety assessment method for the entire driving process of an autonomous vehicle as described in claim 1 includes: The acquisition module is used to acquire first driving data of the target vehicle and second driving data of vehicles in the surrounding area of ​​the target vehicle. The risk assessment module is used to input the first driving data and the second driving data into the risk assessment model to perform a risk assessment on the target vehicle and obtain a risk assessment result; the risk assessment result includes driving risk, collision risk and damage risk. The safety assessment module is used to input the risk assessment results into the safety assessment model to conduct a safety assessment of the target vehicle and obtain the safety assessment results.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images