A parameter calibration method of an automatic driving vehicle driving risk model

Abstract

The present disclosure provides a parameter calibration method of an automatic driving vehicle driving risk model, comprising: designing dynamic traffic scenes containing low, medium and high complexity types, and constructing a dynamic traffic scene database containing traffic element state information; building a driver cognitive risk capturing system for playing videos in the dynamic traffic scene database, recording feedback data of time-varying signal capturing devices for reflecting driver cognitive risk, and aligning the two in time; conducting a driver cognitive risk capturing experiment, using time-synchronized traffic element state information and the feedback data to form an experimental data set; and gradually calibrating the parameters of the driving risk model using low, medium and high complexity scene data by the control variable method. The present disclosure matches dynamic traffic scene information with time-varying signals of driver risk perception, which can improve the accuracy of the driving risk model.

Description

Technical Field

[0001] This disclosure belongs to the field of driving safety technology, and specifically relates to a parameter calibration method for a driving risk model of an autonomous vehicle. Background Technology

[0002] With the rapid development of autonomous driving technology, Level 1 and Level 2 autonomous driving technologies have matured and begun commercial mass production. Research institutions and automakers are now focusing on the research and development of Level 3 and Level 4 advanced autonomous driving technologies and actively promoting open road testing. Intelligent vehicles are gradually becoming known and accepted by the public. In Level 3 and above autonomous driving, the decision-making and control mechanism shifts from the human driver to a computer system, and its driving performance has attracted widespread attention. Especially in terms of traffic safety, the ability to identify road traffic risks and adopt appropriate driving strategies like a human to gain human trust is a key issue that needs to be addressed to achieve large-scale commercial application of advanced autonomous driving.

[0003] Human perception of driving risks has always been a hot topic in the transportation field. Current technologies typically employ pre-set driving risk identification models to provide collision warnings or driving risk information to support autonomous driving decision-making and control. The accuracy of these identifications directly impacts the safe operation of autonomous vehicles. However, driving is an extremely dynamic process; driving risks are closely related to and dynamically change over time, influenced by the vehicle's state, road alignment, and the motion states of other road users. Traditional risk model calibration generally involves task-based or scenario-based driving tests, requiring drivers to complete questionnaires for data analysis and parameter calibration after completing the task. This approach fails to accurately capture the changes in cognitive risk caused by the highly time-varying nature of scenarios, resulting in inaccurate driving risk identification model calibration. This hinders the achievement of near-human-level decision-making and control capabilities in autonomous driving, impacting the development, maturity, and commercial application of autonomous driving technology. Summary of the Invention

[0004] This disclosure aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, the parameter calibration method for the autonomous vehicle driving risk model provided in this disclosure matches dynamic traffic scene information with time-varying signals of driver risk perception, thereby improving the accuracy of the driving risk model. The parameter calibration method includes:

[0006] The design includes dynamic traffic scenarios of low, medium and high complexity. It utilizes driving videos from the driver's perspective recorded by real vehicle road tests or driving simulation equipment to record the operating status of the vehicle and other traffic participants at various times, thereby constructing a dynamic traffic scenario database containing traffic element status information.

[0007] A driver cognitive risk capture system is established, which can play videos in the dynamic traffic scene database, record feedback data from time-varying signal capture devices that reflect driver cognitive risk, and match and align the videos in the dynamic traffic scene video library and the feedback data from the time-varying signal capture devices in time.

[0008] The driver cognitive risk capture system was used to test different drivers in the dynamic traffic scenario, and the time-synchronized traffic element status information and feedback data from the time-varying signal capture device were recorded to form an experimental dataset.

[0009] Using the experimental dataset, we first analyze low-complexity scenario data to calibrate the spatial dimension parameters of the driving risk model; then we analyze medium-complexity scenario data to calibrate the temporal dimension parameters of the driving risk model; finally, we analyze high-complexity scenario data to calibrate the driver attention dimension parameters of the driving risk model.

[0010] In some embodiments, the low-complexity scenario in the dynamic traffic scene includes a stationary vehicle and a moving other traffic participant moving toward or away from the vehicle at a certain speed and angle.

[0011] In some embodiments, the medium-complexity scenario in the dynamic traffic scenario includes a moving vehicle and another moving traffic participant, the vehicle moving forward at a certain speed and the other traffic participant moving toward or away from the vehicle at a certain speed and angle.

[0012] In some embodiments, the high-complexity scenario in the dynamic traffic scenario includes a moving vehicle and multiple moving other traffic participants, the vehicle moving forward at a certain speed and the other traffic participants moving in accordance with traffic rules.

[0013] In some embodiments, the operating status of the vehicle and other surrounding traffic participants includes recording time, vehicle position, vehicle orientation angle, vehicle speed, vehicle acceleration, vehicle type, and the positions, orientation angles, speeds, accelerations, and types of other surrounding traffic participants.

[0014] In some embodiments, the design of the dynamic traffic scenario needs to take into account the different types of traffic participants, initial speeds, and initial positions.

[0015] In some embodiments, the types of traffic participants are categorized as: motor vehicles, non-motor vehicles, pedestrians, and stationary obstacles; wherein, the motor vehicles include: buses, large trucks, lorries, medium-sized passenger vehicles, cars, and motorcycles; the non-motor vehicles include: tricycles and bicycles; stationary obstacles include: traffic cones, lampposts, median strips, guardrails, and curbs; the initial speed of the traffic participants is any speed that complies with road speed limits; the initial position of the traffic participants is any position that complies with road traffic rules.

[0016] In some embodiments, the driver cognitive risk capture system is used to test drivers of different driving experience, genders, and age groups in the dynamic traffic scenario.

[0017] In some embodiments, the parameters of the driving risk model are calibrated using the experimental dataset, including:

[0018] Based on the experimental dataset, the states of the vehicle and other traffic participants at each time point are used as inputs to the driving risk model, the output of the driving risk model is used as the driving risk estimate, and the calculation process data of the driving risk model is retained.

[0019] Using computational process data in low and medium complexity scenarios, and employing the method of controlling variables, a subset of data whose driving risk estimates are only affected by a single parameter to be calibrated is selected. The time and space dimension parameters in the driving risk model are adjusted one by one, so that the driving risk estimates in low and medium complexity scenarios infinitely approach the true values ​​in low and medium complexity scenarios, i.e., the feedback data from time-varying signal capture devices in low and medium complexity scenarios in the experimental dataset.

[0020] By utilizing computational data from high-complexity scenarios, the driver's attention dimension parameter in the driving risk model is adjusted so that the estimated driving risk value in high-complexity scenarios approaches the true value in high-complexity scenarios infinitely, i.e., the feedback signal from the time-varying signal capture device in high-complexity scenarios in the experimental dataset.

[0021] In some embodiments, the driving risk model is:

[0022]

[0023]

[0024]

[0025]

[0026]

[0027] Among them, Rt Let represent the driving risk of the vehicle at time t; i represents the time number in the calculation time domain starting from time t; j represents the numbers of other traffic participants around time t, with a total number of J; G j Let ω represent the total damage to both parties involved in a hypothetical collision between the vehicle and other traffic participants j, calculated based on the vehicle's operating state at time i in the time domain. This damage is defined as the potential collision damage. D ω represents the distance attenuation coefficient. i Indicates the time decay coefficient; α j This represents the contribution weight of the j-th vehicle in the scenario to the final risk, and is the attention dimension parameter of the driver to be calibrated; This represents the shortest distance between the vehicle's profile and other traffic participants j around the i-th time in the computation time domain; This represents the vehicle's velocity at the i-th time point in the computation time domain; This represents the velocity of other traffic participants j around the i-th time in the computation time domain; This represents the angle between the velocity direction of other traffic participants j around the vehicle and the velocity direction of the vehicle at the i-th time in the computation time domain; Indicates the safety characteristic parameters of the vehicle; C represents the safety characteristic parameters of other traffic participants j in the surrounding area; D represents the spatial dimension parameter to be calibrated; and D represents the temporal dimension parameter to be calibrated.

[0028] This disclosure has the following characteristics and beneficial effects:

[0029] This disclosure provides a parameter calibration method for an autonomous vehicle driving risk model. It utilizes a highly sensitive time-varying signal capture device to record the driver's risk perception values ​​in real time and matches them with surrounding vehicle and road information in a traffic scenario, thereby achieving parameter calibration of a human-like cognitive driving risk model. Compared with existing technologies, this disclosure can meet the needs of accurate and highly sensitive capture of driver cognitive risk in dynamic, time-varying traffic scenarios, providing support for research on driver cognitive risk. Attached Figure Description

[0030] Figure 1 This is an overall flowchart of the parameter calibration method for the autonomous vehicle driving risk model provided in the first aspect of this disclosure.

[0031] Figure 2 This is a schematic diagram illustrating a typical operating condition under a low-complexity scenario type designed in the calibration method of the first aspect embodiment of this disclosure.

[0032] Figure 3 This is a schematic diagram illustrating a typical operating condition of the calibration method designed in the first aspect of this disclosure under a medium-complexity scenario.

[0033] Figure 4 This is a schematic diagram illustrating a typical operating condition under a high-complexity scenario type designed in the calibration method of the first aspect embodiment of this disclosure.

[0034] Figure 5 This is a schematic diagram of the structure of a highly sensitive risk-sensing capture system built in the calibration method of the first aspect of this disclosure.

[0035] Figure 6 This is a schematic diagram of the structure of an electronic device provided in a third aspect embodiment of the present disclosure. Specific implementation methods

[0036] To make the objectives, technical solutions, and advantages of this application clearer, the application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for explaining this application and are not intended to limit this application.

[0037] Conversely, this application covers any substitutions, modifications, equivalent methods, and schemes made within the spirit and scope of this application as defined by the claims. Furthermore, to provide the public with a better understanding of this application, certain specific details are described in detail below. However, those skilled in the art will fully understand this application even without these detailed descriptions.

[0038] See Figure 1 The parameter calibration method for an autonomous vehicle driving risk model provided in the first aspect of this disclosure includes:

[0039] S1, designed to include dynamic traffic scenarios of low, medium and high complexity, utilizes driving videos from the driver's perspective recorded by real vehicle road tests or driving simulation equipment to record the operating status of the vehicle and other traffic participants at each moment, thereby constructing a dynamic traffic scenario database containing traffic element status information.

[0040] S2, build a driver cognitive risk capture system. This system can play videos in the dynamic traffic scene database, record feedback data from time-varying signal capture devices that reflect driver cognitive risk, and match and align the videos in the dynamic traffic scene database and the feedback data from time-varying signal capture devices in time.

[0041] S3. Using the driver cognitive risk capture system described above, different drivers are tested in dynamic traffic scenarios. The system records the state information of traffic elements synchronized with time and the feedback data of time-varying signal capture devices to form an experimental dataset.

[0042] S4. Using the experimental dataset mentioned above, we first analyze low-complexity scenario data to calibrate the spatial dimension parameters of the driving risk model; then we analyze medium-complexity scenario data to calibrate the temporal dimension parameters of the driving risk model; finally, we analyze high-complexity scenario data to calibrate the driver's attention dimension parameters of the driving risk model.

[0043] In some embodiments, step S1 specifically includes:

[0044] S11, Design dynamic traffic scenarios of different complexity types. The design process is as follows:

[0045] The low-complexity scenario design consists of only one stationary vehicle and one moving other traffic participant. In different operating conditions under this scenario type, the vehicle is stationary, while the other traffic participants move toward or away from the vehicle at a certain speed and angle. Figure 2 This diagram illustrates typical operating conditions for this scenario type. In scenario 101, the vehicle is stationary. 102 to 106 represent the initial states of other traffic participants under different operating conditions. Arrows indicate the direction of travel. In scenario 102, other traffic participants are in the same lane as the vehicle and approach from behind. In scenario 103, other traffic participants approach and overtake the vehicle in the same direction as it in the right lane. In scenario 104, other traffic participants are traveling in the opposite direction to the vehicle in the left lane. In scenario 105, other traffic participants approach and pass the vehicle from the left at a 90-degree angle. In scenario 106, other traffic participants approach and pass the vehicle from the right front.

[0046] The medium-complexity scenario is designed to include only one moving vehicle and one other moving traffic participant. In different operating conditions under this scenario type, the vehicle moves forward at a certain speed, while the other traffic participants move toward or away from the vehicle at a certain speed and at a certain angle. Figure 3 This demonstrates typical operation in this scenario type, where 201 represents the vehicle itself, moving forward at a constant speed; 202 to 206 represent the initial states of other traffic participants under different operating conditions, similar to cases 102 to 106 in low-complexity scenarios, and will not be elaborated here; the arrows indicate the direction of operation.

[0047] The high-complexity scenario design involves a moving vehicle and multiple other moving traffic participants. In this scenario type, the vehicle moves forward at a certain speed, while the other traffic participants move in accordance with traffic rules. Figure 4 The demonstration shows a typical operation in this scenario, where vehicle 301 is a self-driving vehicle moving forward at a certain speed, while multiple other traffic participants around it simultaneously follow traffic rules and operate autonomously.

[0048] Furthermore, the scenario design needs to consider different types of traffic participants, initial speeds, and initial positions. Traffic participant types can be categorized into four types: motor vehicles, non-motor vehicles, pedestrians, and stationary obstacles. Motor vehicles include buses, large trucks, lorries, medium-sized buses, cars, and motorcycles; non-motor vehicles include tricycles and bicycles; and stationary obstacles include traffic cones, lampposts, medians, guardrails, and curbs. Initial speeds are any speed that complies with road speed limits; initial positions are any positions that comply with road traffic rules.

[0049] In one embodiment, for low-complexity scenarios, for Figure 2 The scenarios 102 to 106 shown are configured with other vehicles in the vicinity as cars, with initial speeds of 20 km / h and 30 km / h, and initial relative positions of 15 m and 25 m, respectively. Using a cross-design approach, there are a total of 2 (other vehicle speed levels) * 2 (distance levels) * 5 (scenarios) = 20 low-complexity scenarios. For... Figure 3 The medium-complexity scenarios shown are designed based on the low-complexity scenario design, with two initial vehicle speeds of 20km / h and 30km / h. Therefore, there are a total of 2 (vehicle speed level) * 2 (other vehicle speed level) * 2 (distance level) * 5 (cases) = 40 medium-complexity scenarios. For high-complexity scenarios, the following approach is adopted... Figure 4 The standard two-way four-lane unsignal-controlled crossroads shown is designed with one other vehicle operating in each entrance lane of the intersection in addition to the vehicle itself, in accordance with traffic rules, for a total of 8 vehicles. All vehicles are cars, with their initial speeds randomly selected between 10km / h and 30km / h, and their initial positions randomly selected between 5m and 20m from the stop line of the intersection. The vehicle's speed is maintained at 20km / h. A total of 5 highly complex traffic scenarios are designed.

[0050] S12, Construct a dynamic traffic scenario database containing traffic element status information:

[0051] After completing the scenario design, the scenario is implemented either in real-world conditions or using driving simulation equipment. During the scenario's operation, a camera is placed at the driver's position to record the driver's first-person driving video. Simultaneously, the driving status of the vehicle and surrounding vehicles at various times is recorded, including data such as position coordinates, speed, orientation, and acceleration. The video and driving status data are then organized and aligned to obtain a dynamic traffic scenario database containing traffic element status information.

[0052] In one embodiment, a driving simulation method was used to implement 20 low-complexity scenarios, 40 medium-complexity scenarios, and 5 high-complexity scenarios, with each scenario having a runtime of 5 seconds. The recorded data includes first-person view video data from the driver, with video files named "102-v20-p15" to represent "situation-speed-position" information, and stored in MP4 format. Simultaneously, the driving status data of the vehicle and surrounding vehicles was recorded at a frequency of 10Hz, generating a 50*n column data matrix, stored in CSV format. Each row of the data matrix contains different fields of data at the same time point, and the columns include: [t, x...]. e ,y e ,φ e ,v e ,a e ,c e ,x s1 ,y s1 ,φ s1 ,v s1 ,a s1 ,c s1 ,x s2 …], t represents the recording time, x e and y e Indicates the position of the vehicle, φ e Indicates the vehicle's heading angle, v e Indicates the vehicle's speed, a e Indicates the vehicle's acceleration, c e Indicates the vehicle type, x s1 and y s1 Indicates the position of the circumferential carriage, φ s1 Indicates the car's heading angle, v s1 Indicates the speed of the chariot, a s1 Represents the acceleration of the circumferential vehicle, c s1 Indicates the type of vehicle, etc.

[0053] In some embodiments, see Figure 5 The driver cognitive risk capture system established in step S2 includes a computer 401 and a time-varying signal capture device 402 with high sensitivity to driver cognitive risk. The time-varying signal capture device 402 is connected to the computer 401 via wired or wireless means for data communication. The computer 401 is used to play the driver's first-person perspective driving video recorded in step S1 and synchronously record the signal from the time-varying signal capture device 402, aligning the signal with the timeline of the driver's first-person perspective driving video.

[0054] In one embodiment, the time-varying signal capture device 402 for driver risk perception uses a Microsoft Surface Dial knob controller. This device can rotate 360 ​​degrees steplessly and has a motion recognition accuracy of 0.1 degrees, meeting the requirements for high-sensitivity feedback. The computer 401 is equipped with experimental software developed using C# language and the API interface provided by Surface Dial, which has video playback and synchronous recording functions for feedback signals, supporting driver risk perception capture experiments.

[0055] In some embodiments, in step S3, in order to ensure the diversity and adaptability of the experimental dataset so as to represent the behavioral habits of drivers in the current traffic environment, tests are conducted on drivers of different driving experience, genders, and age groups.

[0056] In one embodiment, a total of 30 drivers were tested, including 6 women and 24 men; 22 of them had more than 5 years of driving experience. A driver cognitive risk capture experiment was conducted. For each driver, scenarios were randomly ordered and played, and drivers were required to perform actions on the time-varying signal capture device 402 based on their risk perception of the played videos (targeting...). Figure 5 The driver cognitive risk capture system shown is a rotating motion system that generates device signals reflecting the driver's perceived risk. During the experiment, the video number and the signal (perceived risk value) of the time-varying signal capture device were recorded and aligned with the traffic element state information contained in the dynamic traffic scene database on the time axis to form an experimental dataset.

[0057] Specifically, for a driver A, the value p of the knob signal is recorded at a frequency of 10Hz, and time-aligned with the traffic element status information corresponding to the video. This data is then appended to the last column of the existing CSV data file to form a new data file and saved.

[0058] In some embodiments, step S4 specifically includes:

[0059] S41. Based on the collected experimental dataset, the vehicle and surrounding vehicle states at each time point are used as inputs to the driving risk model, the output of the driving risk model is used as the driving risk estimate, and the calculation process data of the driving risk model is retained.

[0060] S42. Using the computation process data of low and medium complexity scenarios obtained in step S42, the method of controlling variables is used to screen the data subset whose driving risk estimate is only affected by a single parameter to be calibrated, and the time and space dimension parameters in the driving risk model are adjusted one by one so that the driving risk estimate output by the driving risk model approaches the true value in low and medium complexity scenarios infinitely, that is, the feedback data from the time-varying signal capture device in the low and medium complexity scenarios in the experimental dataset.

[0061] S43 utilizes computational process data in high-complexity scenarios to adjust the driver's attention dimension parameters in the driving risk model, making the estimated driving risk value in high-complexity scenarios infinitely close to the true value in high-complexity scenarios, i.e., the feedback signal from the time-varying signal capture device in high-complexity scenarios in the experimental dataset.

[0062] In one embodiment, the autonomous vehicle driving risk model to be calibrated is:

[0063]

[0064]

[0065]

[0066]

[0067]

[0068] Among them, R t Let represent the driving risk of the vehicle at time t; i represents the time number in the calculation time domain starting from time t; j represents the number of traffic participants at time t, with a total number of J; G j Let ω represent the total damage to both the vehicle and traffic participant j that would result from a hypothetical collision at the i-th time in the time domain, calculated based on their respective operating states. This damage is defined as the potential collision damage. D ω represents the distance attenuation coefficient. i Indicates the time decay coefficient; α j This represents the contribution weight of the j-th vehicle in the scenario to the final risk, simulating driver attention. This represents the shortest distance between traffic participant j and the vehicle profile at time i in the computation time domain; This represents the vehicle's velocity at the i-th time point in the computation time domain; This represents the velocity of traffic participant j at time i in the computation time domain; This represents the angle between the velocity direction of traffic participant j and the velocity direction of the vehicle at the i-th time in the computation time domain; Indicates the safety characteristic parameters of the vehicle; Let α represent the safety characteristic parameters of traffic participant j; C is the adjustment parameter of the distance decay function, and D is the adjustment parameter of the time decay function. In this embodiment, α j C and D are the parameters to be calibrated in this model.

[0069] S42, using the aforementioned autonomous vehicle driving risk model, firstly, random values ​​are assigned to the parameters to be calibrated to obtain driving risk estimates and calculation process data. The process data includes the collision damage value G at each time point. j Shortest distance d j The time step i corresponding to the shortest distance, etc.

[0070] S43, using the computational process data for low and medium complexity scenarios obtained in step S42, select data with the same collision damage value and the same shortest distance to calibrate the time dimension parameters. In this embodiment, using... Figure 2 Taking scenario 104 as an example, we select two cases with initial speeds of 20 km / h and 30 km / h, and a relative initial position of 15 m. In both cases, the driving risk model exhibits the same collision damage G. j and the same shortest spacing At different times i, i.e., by controlling variables, the estimated driving risk is only related to the time reduction factor ω. i Relatedly, by extracting experimental datasets similar to those described above, the time reduction factor ω can be calculated using the least squares method. i The optimal estimate of the parameter D is obtained. Similarly, using data with the same collision damage value and the time corresponding to the shortest distance, the spatial dimension parameter is calibrated, and the distance attenuation coefficient ω is calculated using the least squares method. D The optimal estimate of parameter C.

[0071] S44. Using the optimal estimates of parameters C and D, calculate the estimated driving risk under high-complexity scenarios, compare it with the true values ​​in the experimental dataset, and calibrate parameter α. j The value of .

[0072] The parameter calibration device for an autonomous vehicle driving risk model provided in the second aspect of this disclosure includes:

[0073] The first module is configured to design dynamic traffic scenarios with three complexity types: low, medium, and high. It uses real vehicle road tests or driving simulation equipment to record driving videos from the driver's perspective and record the operating status of the vehicle and other traffic participants at various times to build a dynamic traffic scenario database containing traffic element status information.

[0074] The driver cognitive risk capture system is used to play videos from a dynamic traffic scene database, record feedback data from time-varying signal capture devices that reflect driver cognitive risk, and match and align the videos from the dynamic traffic scene database and the feedback data from time-varying signal capture devices in time.

[0075] The second module is configured to record the time-synchronized traffic element state information and the feedback data of the time-varying signal capture device when different drivers are tested in dynamic traffic scenarios using the above-mentioned driver cognitive risk capture system, forming an experimental dataset.

[0076] The third module is configured to use the aforementioned experimental dataset to first analyze low-complexity scenario data and calibrate the spatial dimension parameters of the driving risk model; then analyze medium-complexity scenario data and calibrate the temporal dimension parameters of the driving risk model; and finally analyze high-complexity scenario data and calibrate the driver's attention dimension parameters of the driving risk model.

[0077] To implement the above embodiments, this disclosure also proposes a computer-readable storage medium storing a computer program that is executed by a processor to perform the parameter calibration method for the autonomous vehicle driving risk model of the above embodiments.

[0078] The following is for reference. Figure 6 The diagram illustrates a structural schematic of an electronic device suitable for implementing embodiments of the present disclosure. It should be noted that the electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs, desktop computers, and servers. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0079] like Figure 6 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 101, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 102 or a program loaded from a storage device 108 into a random access memory (RAM) 103. The RAM 103 also stores various programs and data required for the operation of the electronic device. The processing unit 101, ROM 102, and RAM 103 are interconnected via a bus 104. An input / output (I / O) interface 105 is also connected to the bus 104.

[0080] Typically, the following devices can be connected to I / O interface 105: input devices 106 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, etc.; output devices 107 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 108 including, for example, magnetic tapes, hard disks, etc.; and communication devices 109. Communication device 109 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0081] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, this embodiment includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network via communication device 109, or installed from storage device 108, or installed from ROM 102. When the computer program is executed by processing device 101, it performs the functions defined above in the methods of embodiments of this disclosure.

[0082] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0083] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0084] The aforementioned computer-readable medium carries one or more programs. When these programs are executed by the electronic device, the electronic device causes the following: It designs dynamic traffic scenarios with low, medium, and high complexity levels; it utilizes driving videos recorded from the driver's perspective based on real-vehicle road tests or simulated driving equipment to record the operating states of the vehicle and other surrounding traffic participants at various times, constructing a dynamic traffic scenario database containing traffic element state information; it records time-synchronized traffic element state information generated by a driver cognitive risk capture system during tests on different drivers in dynamic traffic scenarios, and feedback data from a time-varying signal capture device reflecting driver cognitive risk, forming an experimental dataset. The driver cognitive risk capture system has the function of playing videos from the dynamic traffic scenario database, recording feedback data from the time-varying signal capture device, and matching and aligning the videos and feedback data from the time-varying signal capture device in time; using the aforementioned experimental dataset, it first analyzes low-complexity scenario data to calibrate the spatial dimension parameters of the driving risk model; based on this, it analyzes medium-complexity scenario data to calibrate the temporal dimension parameters of the driving risk model; finally, it analyzes high-complexity scenario data to calibrate the driver's attention dimension parameters of the driving risk model.

[0085] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and Python, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0086] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0087] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0088] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the function involved, as will be understood by those skilled in the art to which embodiments of this application pertain.

[0089] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0090] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0091] Those skilled in the art will understand that implementing all or part of the steps of the methods in the above embodiments can be accomplished by instructing related hardware through a program. The developed program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0092] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0093] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A parameter calibration method for an autonomous vehicle driving risk model, characterized in that, include: The design includes dynamic traffic scenarios of low, medium and high complexity. It utilizes driving videos from the driver's perspective recorded by real vehicle road tests or driving simulation equipment to record the operating status of the vehicle and other traffic participants at various times, thereby constructing a dynamic traffic scenario database containing traffic element status information. A driver cognitive risk capture system is established, which can play videos in the dynamic traffic scene database, record feedback data from time-varying signal capture devices that reflect driver cognitive risk, and match and align the videos in the dynamic traffic scene database and the feedback data from the time-varying signal capture devices in time. The driver cognitive risk capture system was used to test different drivers in the dynamic traffic scenario, and the time-synchronized traffic element status information and feedback data from the time-varying signal capture device were recorded to form an experimental dataset. Using the experimental dataset, we first analyze low-complexity scenario data to calibrate the spatial dimension parameters of the driving risk model; then we analyze medium-complexity scenario data to calibrate the temporal dimension parameters of the driving risk model; finally, we analyze high-complexity scenario data to calibrate the driver attention dimension parameters of the driving risk model. The low-complexity scenario in the dynamic traffic scenario includes a stationary vehicle and another moving traffic participant, which is moving toward or away from the vehicle at a certain speed and angle. The medium-complexity scenario in the dynamic traffic scenario includes a moving vehicle and another moving traffic participant. The vehicle is moving forward at a certain speed, and the other traffic participants are moving toward or away from the vehicle at a certain speed and angle. The highly complex scenario in the dynamic traffic scenario includes a moving vehicle and multiple other moving traffic participants. The vehicle moves forward at a certain speed, while the other traffic participants move in accordance with traffic rules. The driving risk model is as follows: in, express Constant driving risks of own vehicles; Indicated by The time is the time number in the computational time domain, starting from the given time. express The number of other traffic participants in the vicinity at any given time, the total number of which is: ; Indicates the relationship between the vehicle and other road users. To calculate the first in the time domain The total damage to both parties involved in a hypothetical collision at a given moment is defined as potential collision damage. Indicates the distance attenuation coefficient; Indicates the time decay coefficient; Indicates the first in the scene The contribution weight of each vehicle to the final risk is represented by the driver's attention dimension parameter to be calibrated; In the computation time domain, the first... Other traffic participants in the vicinity at that moment The distance between the vehicle's outline and the vehicle's silhouette; In the computation time domain, the first... The speed of the car at that moment; In the computation time domain, the first... Other traffic participants in the vicinity at that moment speed Indicates the safety characteristic parameters of the vehicle; Indicates other traffic participants in the vicinity Safety characteristic parameters; The spatial dimension parameter to be calibrated. The time dimension parameter to be calibrated.

2. The parameter calibration method according to claim 1, characterized in that, The operating status of the vehicle and other traffic participants includes the recording time, vehicle position, vehicle orientation angle, vehicle speed, vehicle acceleration, vehicle type, and the positions, orientation angles, speeds, accelerations, and types of other traffic participants.

3. The parameter calibration method according to claim 1, characterized in that, In designing the dynamic traffic scenario, it is necessary to consider the different types of traffic participants, initial speeds, and initial positions.

4. The parameter calibration method according to claim 3, characterized in that, The types of traffic participants are categorized as follows: motor vehicles, non-motor vehicles, pedestrians, and stationary obstacles. Among them, motor vehicles include buses, trucks, medium-sized passenger vehicles, cars, and motorcycles; non-motor vehicles include tricycles and bicycles; stationary obstacles include traffic cones, lampposts, median strips, guardrails, and curbs; the initial speed of traffic participants is any speed that complies with road speed limits; the initial position of traffic participants is any position that complies with road traffic rules.

5. The parameter calibration method according to claim 1, characterized in that, The driver cognitive risk capture system was used to test drivers of different driving experience, genders, and age groups in the dynamic traffic scenario.

6. The parameter calibration method according to claim 1, characterized in that, The parameters of the driving risk model are calibrated using the experimental dataset, including: Based on the experimental dataset, the states of the vehicle and other traffic participants at each time point are used as inputs to the driving risk model, the output of the driving risk model is used as the driving risk estimate, and the calculation process data of the driving risk model is retained. Using computational process data in low and medium complexity scenarios, and employing the method of controlling variables, a subset of data whose driving risk estimates are only affected by a single parameter to be calibrated is selected. The time and space dimension parameters in the driving risk model are adjusted one by one, so that the driving risk estimates in low and medium complexity scenarios infinitely approach the true values ​​in low and medium complexity scenarios, i.e., the feedback data from time-varying signal capture devices in low and medium complexity scenarios in the experimental dataset. By utilizing computational data from high-complexity scenarios, the driver's attention dimension parameter in the driving risk model is adjusted so that the estimated driving risk value in high-complexity scenarios approaches the true value in high-complexity scenarios infinitely, i.e., the feedback signal from the time-varying signal capture device in high-complexity scenarios in the experimental dataset.

Images