A driving behavior coding and scenario modeling system for unmanned vehicles

Abstract

The present invention relates to the field of intelligent transportation Internet of Things application technology, and specifically discloses a driving behavior coding and scenario modeling system for unmanned vehicles. By comprehensively collecting various parameters of the environment in which the unmanned vehicle operates and accurately evaluating the risk impact of environmental factors on the operation of the unmanned vehicle, a suitable unmanned vehicle behavior reference set is intelligently matched. On this basis, the system further collects performance parameters of the perception layer to which the unmanned vehicle belongs, the vehicle's operating parameters, and the real-time perception parameters of the perception layer. By comprehensively considering the perception layer performance, the vehicle's operating status, and the behavior reference set, the effective behavior factors of the unmanned vehicle can be deeply analyzed and extracted. Based on the comparison between the effective behavior factors of the unmanned vehicle and the effective behavior threshold, the perception parameters of the perception layer of the unmanned vehicle are finely adjusted and encoded, thereby realizing accurate modeling of the unmanned vehicle operation scenario.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation Internet of Things application technology, and specifically to a vehicle driving behavior coding and scene modeling system for unmanned driving. Background Art

[0002] With the continuous advancement of intelligent technology, driverless technology has become an important development direction in the field of transportation. In this field, vehicle driving behavior coding and scenario modeling systems play a vital role. The system combines advanced sensor technology, artificial intelligence algorithms and big data analysis to provide driverless cars with accurate decision support and a reliable operating environment.

[0003] For example, the invention patent with publication number CN118172923A discloses a deep learning modeling method for the interaction between left-turning vehicles and illegal bicycles at intersections, specifically referring to a deep learning model for modeling left-turning vehicle behavior in a mixed flow intersection scenario where there is a potential conflict between vehicles and electric bicycles. The model mainly includes the following steps: preprocessing the data, inputting the vehicle's historical trajectory into the vehicle motion encoder, and extracting the vehicle's historical trajectory features; inputting the relative position from the vehicle to the potential target point into the driving intention extraction module, introducing the maximum pooling layer, and capturing the driving intention features to the potential target point; then inputting the outputs of the first two modules into the target flow interaction module for modeling the interaction between target flow vehicles; inputting the mixed flow object's historical trajectory and the relative position from the vehicle to the mixed flow object into the mixed flow interaction module; and then combining the above extracted features to decode and generate the vehicle's future driving trajectory.

[0004] For example, the invention patent with publication number CN114582142A, a method and system for modeling the vehicle operation status of an intersection with variable lanes, belongs to the field of intelligent transportation technology. When the distance between the vehicle and the opening of the variable lane is greater than a first threshold, the vehicle operation status is obtained according to the straight-line following model; when the distance between the vehicle and the opening of the variable lane is less than the first threshold, whether to enter the variable lane is determined according to the state of the pre-signal light and the lane change selection model; when the vehicle enters the intersection, the operation status of the left-turning vehicle is obtained according to the curve model.

[0005] However, in the process of implementing the embodiments of the present application, the present application discovered that the above technology has at least the following technical problems: when modeling the vehicle driving scene, the above application directly models the collected data, so that the established scene model may not be able to accurately simulate the vehicle's perception behavior and performance under different environmental conditions, resulting in insufficient model generalization ability. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a vehicle driving behavior coding and scene modeling system for unmanned driving, which can effectively solve the problems involved in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a driving behavior coding and scenario modeling system for unmanned vehicles, including an operating environment evaluation module, which is used to collect environmental parameters of the unmanned vehicle, evaluate the risk impact of the operating environment of the unmanned vehicle, and match the behavior reference set of the unmanned vehicle; a data perception analysis module, which is used to collect the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the perception parameters of the perception layer to which the unmanned vehicle belongs, and analyze the behavior effectiveness factor of the unmanned vehicle by integrating the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the behavior reference set of the unmanned vehicle; an operating behavior determination module, which is used to compare the behavior effectiveness factor of the unmanned vehicle with the behavior effectiveness threshold, if the behavior effectiveness factor of the unmanned vehicle is greater than the behavior effectiveness threshold, then the perception parameters of the perception layer to which the unmanned vehicle belongs are behaviorally encoded, if the behavior effectiveness factor of the unmanned vehicle is less than or equal to the behavior effectiveness threshold, then the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and encoded, thereby completing the operation scenario modeling of the unmanned vehicle.

[0008] As a further solution, the behavior reference set of the unmanned vehicle is matched, and the specific matching process is: matching the risk impact of the environment in which the unmanned vehicle operates within the first monitoring cycle with the behavior reference set corresponding to each risk impact interval stored in the coding database, thereby obtaining the behavior reference set of the unmanned vehicle.

[0009] As a further solution, the analysis of the effective behavioral factors of the unmanned vehicle specifically refers to the analysis of the effective behavioral factors of the unmanned vehicle within the first monitoring period, and the specific analysis method is: comprehensively analyze the performance parameters of the perception layer to which the unmanned vehicle belongs during the first monitoring period and the behavior reference set of the unmanned vehicle, and analyze the effective perception factors of the unmanned vehicle during the first monitoring period; comprehensively analyze the operating parameters of the unmanned vehicle during the first monitoring period and the behavior reference set of the unmanned vehicle, and analyze the effective operation factors of the unmanned vehicle during the first monitoring period; comprehensively process the perception effective factors of the unmanned vehicle during the first monitoring period, the operating effective factors of the unmanned vehicle during the first monitoring period, and the risk impact of the environment to which the unmanned vehicle operates during the first monitoring period, to obtain an analysis method for the effective behavioral factors of the unmanned vehicle during the first monitoring period.

[0010] As a further solution, the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and then encoded. The specific adjustment process is: the behavior effectiveness factor of the unmanned vehicle in the first monitoring period is ratioed with the behavior effectiveness threshold to obtain the behavior effectiveness ratio of the unmanned vehicle in the first monitoring period, and the ratio is matched with the adjustment set corresponding to each behavior effectiveness ratio interval stored in the coding database to obtain the adjustment set of the unmanned vehicle in the first monitoring period; the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted through the adjustment set and re-encoded.

[0011] As a further solution, the operation scenario modeling of the unmanned vehicle is completed, and the specific process is: obtain the operation process of the unmanned vehicle, divide it into several monitoring periods according to equal time intervals, and extract the first monitoring period by time sequence numbering, thereby encoding the perception parameters of the perception layer to which the unmanned vehicle belongs within the first monitoring period; through the above process, the perception parameters of the perception layer to which the unmanned vehicle belongs within several monitoring periods are encoded and integrated, and the integrated code is marked as the total code of the operation scenario of the unmanned vehicle; the total code of the operation scenario of the unmanned vehicle is decoded to complete the operation scenario modeling of the unmanned vehicle.

[0012] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0013] (1) The present invention provides a vehicle driving behavior coding and scenario modeling system for unmanned driving. The system comprehensively collects various parameters of the environment in which the unmanned vehicle operates and accurately evaluates the risk impact of environmental factors on the operation of the unmanned vehicle, thereby intelligently matching a suitable unmanned vehicle behavior reference set. On this basis, the system further collects performance parameters of the perception layer to which the unmanned vehicle belongs, vehicle operation parameters, and real-time perception parameters of the perception layer. By comprehensively considering the performance of the perception layer, the vehicle operation status, and the behavior reference set, the system can deeply analyze and extract the effective behavior factors of the unmanned vehicle. Based on the comparison between the effective behavior factors of the unmanned vehicle and the effective behavior threshold, the perception parameters of the perception layer of the unmanned vehicle are finely adjusted and encoded, thereby realizing accurate modeling of the unmanned vehicle operation scenario. This not only significantly improves the accuracy of vehicle driving behavior coding and scenario modeling, but also improves the environmental adaptability and decision-making accuracy of the unmanned vehicle, and enhances its safety and reliability in complex scenarios.

[0014] (2) The present invention analyzes the effective behavioral factors of unmanned vehicles by comprehensively analyzing the performance parameters of the perception layer to which the unmanned vehicles belong, the operating parameters of the unmanned vehicles, and the behavioral reference set of the unmanned vehicles. This comprehensive analysis method effectively reduces the limitations of single-dimensional analysis and ensures the comprehensiveness and accuracy of the analysis of the behavioral decisions of unmanned vehicles. Through this method, unmanned vehicles can more accurately understand and respond to complex and changing traffic environments, significantly improving their driving safety and efficiency. At the same time, it also provides solid and scientific data support and basis for the parameters perceived by unmanned vehicles, promoting the overall progress of intelligent transportation systems.

[0015] (3) The present invention compares the behavior effectiveness factor of the unmanned vehicle with the behavior effectiveness threshold. If the behavior effectiveness factor of the unmanned vehicle is less than or equal to the behavior effectiveness threshold, the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and encoded. This adjustment mechanism ensures that the unmanned vehicle can make more reasonable and effective decisions when facing a complex environment, thereby ensuring the accuracy of the unmanned vehicle's driving behavior encoding and scene modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0017] Figure 1 It is a schematic diagram of system module connection of the present invention.

[0018] Figure 2 This is a schematic diagram of adjacent vehicles involved in the present invention.

[0019] Figure numerals: 1. Unmanned vehicle; 2. Adjacent vehicle; 3. Ordinary vehicle; 4. Sensing area. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0021] See also Figure 1 As shown, an embodiment of the present invention provides a technical solution: a vehicle driving behavior coding and scenario modeling system for unmanned driving, including an operating environment assessment module, a data perception analysis module and an operating behavior determination module.

[0022] The operating environment assessment module is connected to the data perception analysis module, the data perception analysis module is connected to the operating behavior determination module, and the operating environment assessment module, the data perception analysis module and the operating behavior determination module are all connected to the coding database.

[0023] The operating environment assessment module is used to collect parameters of the environment in which the unmanned vehicle operates, assess the risk impact of the environment in which the unmanned vehicle operates, and match a behavior reference set of the unmanned vehicle.

[0024] Specifically, the behavior reference set of the unmanned vehicle is matched, and the specific matching process is as follows:

[0025] The risk impact of the environment in which the unmanned vehicle operates in the first monitoring cycle is matched with the behavior reference sets corresponding to each risk impact interval stored in the coding database, thereby obtaining the behavior reference set of the unmanned vehicle; in an example embodiment, assuming that the risk impact of the environment in which the unmanned vehicle operates is X, in the risk impact interval [X-45%, X+35%] stored in the coding database, the behavior reference set corresponding to the risk impact interval [X-45%, X+35%] includes: a minimum perception area limit value of 35 square meters, a data refresh average interval time limit value of 0.8 seconds, an object tracking frame reference average ratio of 5%, an instruction response average limit time of 0.3 seconds, a yaw rate limit value of 8% and a total angular deviation limit value of 180 degrees.

[0026] It should be explained that the behavior reference set includes but is not limited to the minimum perception area limit value, the average data refresh interval time limit value, the object tracking frame reference average ratio, the command response average time limit value, the yaw rate limit value and the total angular deviation limit value. In view of the fact that the specific application scenarios and technical requirements of this embodiment are closely related to the above-mentioned parameters, the above-mentioned parameters are specially selected for detailed explanation and setting.

[0027] Furthermore, the evaluation of the risk impact of the environment in which the unmanned vehicle operates specifically refers to the risk impact of the environment in which the unmanned vehicle operates within the first monitoring cycle, and the specific evaluation process is as follows:

[0028] The traffic density of the environment in which the unmanned vehicle operates in the first monitoring cycle and the minimum distance between the unmanned vehicle and each adjacent vehicle in the first monitoring cycle are extracted from the parameters of the environment in which the unmanned vehicle operates. The minimum distances between the unmanned vehicle and each adjacent vehicle in the first monitoring cycle are sorted in ascending order, and the minimum distance ranked first is extracted and marked as the minimum distance between the unmanned vehicle in the first monitoring cycle; the above-mentioned first monitoring cycle refers to the first monitoring period divided by equal time intervals preset by the data analyst during the operation of the unmanned vehicle, and this period is the starting part of the entire operation sequence; the above-mentioned adjacent vehicles refer to vehicles in the perception area covered by the perception layer of the unmanned vehicle, specifically Figure 2 As shown in the schematic diagram of adjacent vehicles, each vehicle 2 is marked as each adjacent vehicle 2 of the unmanned vehicle in the perception area 4 covered by the perception layer of the unmanned vehicle 1, and the ordinary vehicle 3 does not belong to the perception area 4 covered by the perception layer of the unmanned vehicle, and therefore does not belong to the adjacent vehicle of the unmanned vehicle; the traffic density of the environment in which the above-mentioned unmanned vehicle operates in the first monitoring period, using the perception layer of the unmanned vehicle to detect and record in real time the vehicle information passing through the perception area in the first monitoring period and its passing time, clarify the specific duration of the first monitoring period and the length of the road monitored by the unmanned vehicle, count the total number of vehicles passing through the monitored road in the first monitoring period, and according to the definition of traffic density (the number of vehicles passing through the unit length of the road), the traffic density is obtained by the calculation formula; the minimum vehicle distance between the above-mentioned unmanned vehicle and each adjacent vehicle in the first monitoring period, the real-time vehicle distance between the unmanned vehicle and each adjacent vehicle in the first monitoring period is obtained through the perception layer of the unmanned vehicle (such as a laser radar), and they are sorted in order from small to large, and the minimum vehicle distance ranked first is extracted, thereby obtaining the minimum vehicle distance between the unmanned vehicle and each adjacent vehicle in the first monitoring period.

[0029] The average speed of each sampled vehicle in the environment where the unmanned vehicle operates within the first monitoring period is extracted from the parameters of the environment where the unmanned vehicle operates, and standard deviation processing is performed, and the processing result is marked as the discrete value of the traffic speed in the environment where the unmanned vehicle operates within the first monitoring period. The above-mentioned sampled vehicles refer to representative vehicles in the environment where the unmanned vehicle operates that are selected for analysis within the first monitoring period. These vehicles are obtained by systematically sampling all vehicles in the environment where the unmanned vehicle operates (for example, vehicles are extracted from the environment at predetermined time intervals or spatial intervals), and the purpose is to reflect the traffic conditions of the entire environment through them; the average speed of each sampled vehicle in the environment where the unmanned vehicle operates within the first monitoring period can be measured by an external traffic monitoring system (such as an intelligent traffic sensor network).

[0030] Thus, a method for evaluating the risk impact of the environment in which the unmanned vehicle operates is comprehensively obtained. In this embodiment, the risk impact of the environment in which the unmanned vehicle operates is obtained by comprehensively analyzing the traffic density, minimum vehicle distance and traffic speed discrete values ​​of the environment in which the unmanned vehicle operates in the first monitoring period, and is used to determine the value of the impact of the environment on the safety risk of the unmanned vehicle operation. The specific evaluation method is:

[0031] ;

[0032] It is the traffic density of the environment in which the unmanned vehicle operates in the first monitoring cycle, which refers to the instantaneous number of vehicles on a unit length section of the road in which the unmanned vehicle operates. It is used to indicate the density of vehicles on a road. In the environment in which the unmanned vehicle operates, traffic density is an important indicator for measuring traffic flow and road congestion.

[0033] It is the minimum vehicle distance of the unmanned vehicle in the first monitoring cycle, which refers to the minimum safe distance maintained between the unmanned vehicle and the surrounding adjacent vehicles in the first monitoring cycle.

[0034] It is the discrete value of the traffic speed in the environment where the unmanned vehicle operates in the first monitoring cycle. It refers to the discrete degree of speed distribution of each sampled vehicle in the environment where the unmanned vehicle operates in the first monitoring cycle. It is an important indicator for evaluating traffic flow stability and safety. When the discrete value of traffic speed is small, it indicates that the speed difference of vehicles on the road is small, the traffic density may be large, and there is a greater safety hazard.

[0035] e is a natural constant, It is the traffic density limit value preset in the coding database, which represents the maximum allowable traffic density of the environment in which the unmanned vehicle operates during the first monitoring cycle.

[0036] It is a minimum vehicle distance limit value preset in the coding database, indicating the minimum allowable minimum vehicle distance of the unmanned vehicle in the first monitoring cycle.

[0037] It is a preset discrete value for defining the traffic speed in the coding database, indicating the minimum allowable discrete value of the traffic speed in the environment in which the unmanned vehicle operates within the first monitoring cycle.

[0038] in, is the risk impact of the environment in which the unmanned vehicle operates in the first monitoring cycle. When the traffic density is large, the number of vehicles on the road increases significantly, resulting in a shortened relative distance between vehicles, that is, a decrease in the minimum vehicle distance. In this scenario, the degree of proximity between vehicles increases. At the same time, the increase in traffic density may also lead to a decrease in the discrete value of traffic speed, indicating that the speed difference between vehicles becomes less obvious and there is a possibility of congestion. This speed difference not only increases the complexity of relative motion between vehicles, but also further increases the possibility of collision. The above parameters jointly reflect the risk level of the environment in which the unmanned vehicle operates. Therefore, in this high-risk simulation environment, matching the behavior reference set of the unmanned vehicle can not only effectively constrain the operation status of the unmanned vehicle and ensure that it complies with the established safety specifications and driving logic, but also provide a reliable judgment basis for the unmanned vehicle during operation. Through the behavior reference set, the unmanned vehicle can more accurately understand and respond to various complex traffic scenarios, and achieve more accurate and efficient driving behavior coding.

[0039] In this example embodiment, the risk impact of the environment in which the unmanned vehicle operates and the change table of its corresponding parameters are shown in Table 1:

[0040] Table 1 Risk impact of the environment in which unmanned vehicles operate and changes in their corresponding parameters

[0041] Traffic density / vehicles / km Minimum distance between vehicles / meter Traffic speed discrete value / m / s Risk impact / percentage 12 4.5 7 219 15 3.7 6.2 223 20 2.1 5 231

[0042] In this example embodiment, the traffic density limit is set to 25 vehicles / km, the minimum vehicle distance limit is set to 1.5 meters, and the traffic speed discrete value is set to 8 meters / second. As can be seen from Table 1, when the traffic density is large (20 vehicles / km), the minimum vehicle distance decreases accordingly (to 2.1 meters), and the traffic speed shows a significant discrete characteristic, and the degree of deviation from the average value is large, which is specifically manifested in that the traffic speed discrete value is large (to 5 meters / second), which has a comprehensive effect on the risk impact, causing the risk impact to increase (to 231%).

[0043] The data perception analysis module is used to collect the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the perception parameters of the perception layer to which the unmanned vehicle belongs, and analyzes the effective factors of the behavior of the unmanned vehicle by integrating the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the behavior reference set of the unmanned vehicle.

[0044] In a specific embodiment, the present invention analyzes the effective behavioral factors of the unmanned vehicle by comprehensively analyzing the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the behavioral reference set of the unmanned vehicle. This comprehensive analysis method effectively reduces the limitations of single-dimensional analysis and ensures the comprehensiveness and accuracy of the analysis of the behavioral decisions of the unmanned vehicle. Through this method, the unmanned vehicle can more accurately understand and respond to the complex and changeable traffic environment, significantly improving its driving safety and efficiency. At the same time, it also provides solid and scientific data support and basis for the parameters perceived by the unmanned vehicle, promoting the overall progress of the intelligent transportation system.

[0045] Specifically, the collecting of the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle and the perception parameters of the perception layer to which the unmanned vehicle belongs refers to collecting the performance parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period, the operating parameters of the unmanned vehicle in the first monitoring period and the perception parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period; the performance parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period include the minimum perception area, the average interval time of data refresh, the total number of tracking frames and the number of tracking frames to which each sampled reference object belongs; the operating parameters of the unmanned vehicle in the first monitoring period include the average time of command response, the operating trajectory curve and the direction angle; the perception parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period include the road type (such as expressways, urban roads and rural roads, etc.), the number of lanes, the lane width, the road curvature, the slope, the instantaneous speed, the average speed and the acceleration of the vehicle, the fixed obstacles on the road (such as construction fences and roadblocks, etc.) and the dynamic obstacles (such as faulty vehicles and scattered objects, etc.). In addition to the above parameters, this example can also be used for other parameters.

[0046] Specifically, the analysis of the effective factors of the behavior of the unmanned vehicle specifically refers to the analysis of the effective factors of the behavior of the unmanned vehicle in the first monitoring period, and the specific analysis method is: comprehensively analyze the performance parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period and the behavior reference set of the unmanned vehicle, and analyze the effective factors of the perception of the unmanned vehicle in the first monitoring period; comprehensively analyze the operating parameters of the unmanned vehicle in the first monitoring period and the behavior reference set of the unmanned vehicle, and analyze the effective factors of the operation of the unmanned vehicle in the first monitoring period; perform comprehensive data processing on the effective factors of the perception of the unmanned vehicle in the first monitoring period, the effective factors of the operation of the unmanned vehicle in the first monitoring period, and the risk impact of the environment to which the unmanned vehicle operates in the first monitoring period, to obtain the analysis method of the effective factors of the behavior of the unmanned vehicle in the first monitoring period; specifically, the specific analysis process of the effective factors of the perception of the unmanned vehicle in the first monitoring period is: from the unmanned vehicle The minimum perception area of ​​the perception layer of the unmanned vehicle in the first monitoring cycle and the average interval duration of data refresh of the perception layer of the unmanned vehicle in the first monitoring cycle are extracted from the performance parameters of the perception layer of the unmanned vehicle in the first monitoring cycle; the minimum perception area of ​​the perception layer of the unmanned vehicle in the first monitoring cycle refers to sorting the real-time perception areas of the perception layer of the unmanned vehicle in the first monitoring cycle in order from small to large, extracting the perception area ranked first, and marking it as the minimum perception area of ​​the perception layer of the unmanned vehicle in the first monitoring cycle; the average interval duration of data refresh of the perception layer of the unmanned vehicle in the first monitoring cycle is obtained by analyzing the perception system log of the unmanned vehicle, obtaining the timestamp of each perception data update, calculating the time interval between two consecutive perception data updates according to the obtained timestamp, and taking the average value, thereby obtaining the average interval duration of data refresh of the perception layer of the unmanned vehicle in the first monitoring cycle.

[0047] The total number of tracking frames of the perception layer of the unmanned vehicle in the first monitoring period and the number of tracking frames of each sampled reference object are extracted from the performance parameters of the perception layer of the unmanned vehicle in the first monitoring period, the number of tracking frames of each sampled reference object is ratioed to the total number of tracking frames, and the results of the ratio processing are averaged, thereby marking the average processing result as the average ratio of the object tracking frames of the perception layer of the unmanned vehicle in the first monitoring period; the total number of tracking frames of the perception layer of the unmanned vehicle in the first monitoring period is specifically obtained by: extracting the sampling rate of the unmanned vehicle from the system configuration of the perception layer of the unmanned vehicle, and the sampling rate refers to the number of frames that the perception system can collect per second. The number of data frames is obtained, and then the duration corresponding to the first monitoring cycle is obtained. The total number of tracking frames of the perception layer to which the unmanned vehicle belongs in the first monitoring cycle is obtained by the total number of tracking frames = the duration corresponding to the first monitoring cycle (seconds) × sampling rate (frames / second); the number of tracking frames to which the above-mentioned sampled reference objects belong is specifically obtained as follows: clearly select the sampled reference objects, which can be vehicles, pedestrians and non-motor vehicles on the road, etc., and apply tracking algorithms (such as Meanshift, Camshift and MOSSE, etc.) to continuously track the marked sampled reference objects, and record the number of frames tracked for each sampled reference object. For example, the number of tracking frames for sampled vehicle A is 100 frames, and the number of tracking frames for sampled vehicle B is 120 frames.

[0048] An analysis method for extracting a minimum perception area boundary value, a data refresh average interval duration boundary value, and an object tracking frame reference average ratio from an unmanned vehicle behavior reference set, thereby comprehensively obtaining a perception effectiveness factor of the unmanned vehicle in the first monitoring period. In this embodiment, the perception effectiveness factor of the unmanned vehicle in the first monitoring period is obtained by comprehensively analyzing the minimum perception area, the data refresh average interval duration, and the object tracking frame average ratio of the perception layer to which the unmanned vehicle belongs in the first monitoring period, and is used to analyze the quantitative index of the working efficiency of the perception layer of the unmanned vehicle in the first monitoring period. The specific analysis method is as follows: ;

[0049] It is the minimum perception area of ​​the perception layer of the unmanned vehicle during the first monitoring cycle, which refers to the minimum geographical area within which the perception layer of the unmanned vehicle can stably and accurately capture and process information during the first monitoring cycle.

[0050] It is the average interval duration of data refresh of the perception layer of the unmanned vehicle in the first monitoring cycle, which refers to the average time difference between two consecutive updates of the perception data of the perception layer of the unmanned vehicle in the first monitoring cycle.

[0051] It is the average ratio of object tracking frames of the perception layer of the unmanned vehicle in the first monitoring cycle, which refers to the average proportion of the number of frames in which the perception layer of the unmanned vehicle successfully tracks the sampled objects (such as vehicles and pedestrians) in the first monitoring cycle to the total number of frames.

[0052] e is a natural constant, It is the minimum perception area limit value, which represents the minimum allowable value of the minimum perception area of ​​the perception layer to which the unmanned vehicle belongs in the first monitoring cycle.

[0053] The average interval length limit for data refresh is the minimum allowable value for the average interval length for data refresh of the perception layer of the unmanned vehicle in the first monitoring cycle.

[0054] is the reference average ratio of object tracking frames, which represents the reference value of the average ratio of object tracking frames of the perception layer to which the unmanned vehicle belongs in the first monitoring cycle.

[0055] in, is the perception effectiveness factor of the unmanned vehicle in the first monitoring cycle. The size of the minimum perception area will affect the average interval time of the unmanned vehicle perception data refresh and the average ratio of the object tracking frame. When the perception area increases, the amount of information that the perception layer needs to process will also increase, which should lead to a longer data refresh interval, but at the same time it will also improve the tracking ability of moving objects. Therefore, when the average interval of data refresh is too long, even if the minimum perception area is large enough, the perception layer cannot reflect environmental changes in real time, which means that the vehicle may not be able to respond in time when the surrounding situation changes. If the average ratio of object tracking frames is low, even if the perception area and data refresh interval are ideal, the perception layer cannot effectively identify and track dynamic targets in the environment, which may cause the unmanned vehicle to be unable to avoid obstacles or predict the behavior of other vehicles in time during driving, thereby reducing driving safety and efficiency, and further causing the unmanned vehicle driving behavior coding and scene modeling to be not based on the optimal operating state. Therefore, a comprehensive analysis of the above parameters can optimize the perception layer performance of the unmanned vehicle, enabling it to more accurately capture and interpret the information of the unmanned vehicle's surrounding environment, and further promote the effectiveness and accuracy of vehicle driving behavior coding and scene modeling.

[0056] Specifically, the operation effectiveness factor of the unmanned vehicle in the first monitoring period is analyzed in the following process:

[0057] The average duration of the command response of the unmanned vehicle in the first monitoring period and the operation trajectory curve of the unmanned vehicle in the first monitoring period are extracted from the operation parameters of the unmanned vehicle in the first monitoring period, the expected operation trajectory curve of the unmanned vehicle in the first monitoring period is obtained, and the operation trajectory curve of the unmanned vehicle in the first monitoring period is compared with the expected operation trajectory curve of the unmanned vehicle in the first monitoring period for curve overlap, thereby obtaining the yaw rate of the unmanned vehicle in the first monitoring period through data analysis; the average duration of the command response of the unmanned vehicle in the first monitoring period is recorded by the unmanned vehicle control system for each command response of the unmanned vehicle in the first monitoring period, and the average duration of the command response of the unmanned vehicle in the first monitoring period is obtained by averaging, wherein the commands include but are not limited to acceleration commands command, deceleration command, steering command, lane change command and parking command; the running trajectory curve of the above-mentioned unmanned vehicle in the first monitoring period is extracted from the data collected by the sensors of the unmanned vehicle (such as radar, camera and lidar, etc.); the expected running trajectory curve of the above-mentioned unmanned vehicle in the first monitoring period is prepared in advance by the modeling engineer according to the actual situation; the yaw rate of the above-mentioned unmanned vehicle in the first monitoring period, the specific data analysis process is: the running trajectory curve of the unmanned vehicle in the first monitoring period and the expected running trajectory curve of the unmanned vehicle in the first monitoring period are compared by using the Hausdorff distance algorithm, etc., and the Hausdorff distance algorithm will output the degree of deviation between the two running trajectory curves and mark it as the yaw rate of the unmanned vehicle in the first monitoring period.

[0058] The direction angle of the unmanned vehicle at each detection time point is extracted from the operating parameters of the unmanned vehicle in the first monitoring cycle, and the expected direction angle of the unmanned vehicle at each detection time point is extracted from the behavior reference set of the unmanned vehicle, and the difference processing is performed with the direction angle of the unmanned vehicle at the corresponding detection time point, and the difference processing is performed on the absolute value, and the results of the absolute value processing are accumulated and marked as the total deviation value of the direction angle of the unmanned vehicle in the first monitoring cycle; the above-mentioned detection time points refer to a number of time points randomly arranged in the first monitoring cycle; the direction angle of the unmanned vehicle at each detection time point is measured by the inertial measurement unit of the unmanned vehicle; the expected direction angle of the unmanned vehicle at each detection time point is extracted from the behavior reference set of the unmanned vehicle, and is prepared in advance by the modeling engineer according to the actual situation; in an example embodiment, the left direction angle is marked as a positive angle, and the right direction angle is marked as a negative angle. Assuming that the direction angle of the unmanned vehicle at a certain detection time point is 10 degrees to the left, and the expected direction angle of the unmanned vehicle at the detection time point is 10 degrees to the right, then the deviation value of the direction angle at this time point is 20 degrees.

[0059] At the same time, the average command response limit time, the yaw rate limit value and the azimuth total deviation limit value are extracted from the behavior reference set of the unmanned vehicle, and the analysis method of the operation effectiveness factor of the unmanned vehicle in the first monitoring period is comprehensively obtained. In this embodiment, the above-mentioned operation effectiveness factor of the unmanned vehicle in the first monitoring period is obtained by comprehensively analyzing the average command response time, yaw rate and azimuth total deviation value of the unmanned vehicle in the first monitoring period, and is used to evaluate the operation efficiency and stability of the unmanned vehicle in the first monitoring period. The specific analysis method is as follows: ;

[0060] It is the average response time of the command of the unmanned vehicle in the first monitoring cycle, which refers to the average time required for the unmanned vehicle to receive the command and start executing the command in the first monitoring cycle.

[0061] It is the yaw rate of the unmanned vehicle in the first monitoring cycle, which refers to the degree of deviation between the actual driving trajectory of the unmanned vehicle in the first monitoring cycle and the expected driving trajectory.

[0062] It is the total deviation value of the direction angle of the unmanned vehicle in the first monitoring cycle, which refers to the total deviation between the direction angle and the expected direction angle during the actual driving process of the unmanned vehicle in the first monitoring cycle.

[0063] e is a natural constant, The average command response time is defined as the maximum allowable average command response time of the unmanned vehicle in the first monitoring cycle.

[0064] is a yaw rate limit value, which indicates the maximum allowable yaw rate of the unmanned vehicle in the first monitoring cycle.

[0065] is the total angular deviation limit value, which indicates the maximum allowable total angular deviation value of the unmanned vehicle in the first monitoring cycle.

[0066] in, It is the effective factor of the operation of the unmanned vehicle in the first monitoring cycle. The longer the average command response time is, the slower the unmanned vehicle reacts to the control command, which may cause the unmanned vehicle to deviate from the predetermined path during driving, that is, the yaw rate increases. At the same time, the longer the average command response time is, the more angular deviations that may be generated by the unmanned vehicle when executing the command, resulting in an increase in the total angular deviation value. The yaw rate reflects the degree to which the unmanned vehicle deviates from the predetermined path, and the total angular deviation value is the specific manifestation of this deviation in the angular direction. In summary, the average command response time and the yaw rate There is a mutual influence between the total angular deviation value and the slow command response, the increase in yaw rate and the accumulation of angular deviation. These changes not only reduce the safety and efficiency of the unmanned vehicle during driving, but also further affect the accuracy of its behavior coding and scene modeling, because behavior coding and scene modeling are based on the unmanned vehicle's perception and response to the environment. When the response of the unmanned vehicle becomes slow and the path tracking ability decreases, the accuracy of the model constructed by the unmanned vehicle will decrease, which may lead to deviations in the decision-making and actions of the unmanned vehicle in complex environments.

[0067] Furthermore, the behavior effectiveness factor of the unmanned vehicle in the first monitoring period, in this embodiment, is obtained by comprehensively analyzing the perception effectiveness factor, operation effectiveness factor and risk impact of the environment in which the unmanned vehicle operates in the first monitoring period, and is used to determine the value of the effectiveness of the behavior performance of the unmanned vehicle in the first monitoring period on the vehicle driving behavior encoding and scenario modeling. The specific analysis method is:

[0068] ;

[0069] It is the perception effectiveness factor of the unmanned vehicle in the first monitoring cycle. The higher the perception effectiveness factor, the more accurately and timely the perception system of the unmanned vehicle can identify and understand the surrounding environment, thereby making more reasonable driving decisions. On the contrary, when the perception effectiveness factor is low, the unmanned vehicle may not be able to identify important information such as obstacles and traffic signals in time, resulting in safety hazards in driving behavior, further reducing the effectiveness of vehicle driving behavior coding and scenario modeling.

[0070] It is the operational effectiveness factor of the unmanned vehicle in the first monitoring cycle. The higher the operational effectiveness factor, the more accurate and stable the unmanned vehicle is in executing driving decisions, and it can better adapt to complex road conditions and ensure driving safety. On the contrary, when the operational effectiveness factor is low, the unmanned vehicle may not be able to accurately execute driving decisions, resulting in unstable vehicle movement, increasing the risk of traffic accidents, and further reducing the effectiveness of vehicle driving behavior coding and scenario modeling.

[0071] It is the risk impact of the operating environment of the unmanned vehicle in the first monitoring cycle. The higher the risk impact of the operating environment, the greater the potential risk faced by the unmanned vehicle, and more cautious driving is required to ensure safety. On the contrary, when the risk impact of the operating environment is low, the unmanned vehicle can drive in a relatively safe environment, thereby improving the effectiveness of vehicle driving behavior coding and scenario modeling.

[0072] e is a natural constant, It is the weight factor corresponding to the perception effectiveness factor preset in the coding database, which indicates the proportion of the perception effectiveness factor to the behavior effectiveness factor of the unmanned vehicle in the first monitoring period. When in use, the weight factor corresponding to the perception effectiveness factor can be directly obtained from the coding database, and the corresponding relationship can be a preset mapping relationship. For example, the minimum perception area, the average interval duration of data refresh, and the average ratio of object tracking frames form a mapping set with the weight factor corresponding to the perception effectiveness factor preset in the coding database. The real-time minimum perception area, the average interval duration of data refresh, and the average ratio of object tracking frames are input into the mapping set to obtain the weight factor corresponding to the perception effectiveness factor. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0073] It is the weight factor corresponding to the operation effectiveness factor preset in the coding database, which indicates the proportion of the operation effectiveness factor to the behavior effectiveness factor of the unmanned vehicle in the first monitoring cycle. When in use, the weight factor corresponding to the operation effectiveness factor can be directly obtained from the coding database, and the corresponding relationship can be a preset mapping relationship. For example, the average command response time, yaw rate and total angular deviation value form a mapping set with the weight factor corresponding to the operation effectiveness factor preset in the coding database. The real-time average command response time, yaw rate and total angular deviation value are input into the mapping set to obtain the weight factor corresponding to the operation effectiveness factor. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0074] It is the weight factor corresponding to the risk impact degree preset in the coding database, which indicates the proportion of the risk impact degree to the effective factors of the behavior of the unmanned vehicle in the first monitoring period. When used, the weight factor corresponding to the risk impact degree can be directly obtained from the coding database, and the corresponding relationship can be a preset mapping relationship. For example, the traffic density, the minimum vehicle distance and the traffic speed discrete values ​​form a mapping set with the weight factors corresponding to the risk impact degree preset in the coding database. The real-time traffic density, the minimum vehicle distance and the traffic speed discrete values ​​are input into the mapping set to obtain the weight factor corresponding to the risk impact degree. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0075] in, It is the effective factor of the behavior of the unmanned vehicle in the first monitoring cycle. When the impact of environmental risks is higher, the information that the unmanned vehicle needs to perceive is more complex and changeable. This requires the perception layer of the unmanned vehicle to have higher sensitivity and accuracy to deal with various potential risks. The impact of environmental risks will also limit the effective factor of the operation of the unmanned vehicle. At the same time, when the environmental risk is high, the vehicle may need to adopt a more conservative driving operation strategy to ensure driving safety and improve the accuracy of vehicle driving behavior coding and scene modeling. Therefore, by analyzing the above parameters, it not only reveals the actual performance of the unmanned vehicle in the current environment, but also provides valuable data support for further optimization and improvement of driving behavior coding. Using the analysis results of these parameters, the perception system, operation strategy and scene modeling process of the unmanned vehicle can be adjusted in a targeted manner.

[0076] The operation behavior determination module is used to compare the behavior effectiveness factor of the unmanned vehicle with the behavior effectiveness threshold. If the behavior effectiveness factor of the unmanned vehicle is greater than the behavior effectiveness threshold, the perception parameters of the perception layer to which the unmanned vehicle belongs are behavior encoded. If the behavior effectiveness factor of the unmanned vehicle is less than or equal to the behavior effectiveness threshold, the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and encoded, thereby completing the modeling of the operation scenario of the unmanned vehicle.

[0077] The above-mentioned behavior validity threshold represents the minimum value of the reasonable range of the behavior validity factor of the unmanned vehicle, which is extracted from the coding database.

[0078] In a specific embodiment, the present invention compares the behavior effectiveness factor of the unmanned vehicle with the behavior effectiveness threshold. If the behavior effectiveness factor of the unmanned vehicle is less than or equal to the behavior effectiveness threshold, the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and encoded. This adjustment mechanism ensures that the unmanned vehicle can make more reasonable and effective decisions when facing a complex environment, thereby ensuring the accuracy of the unmanned vehicle's driving behavior coding and scene modeling.

[0079] Specifically, the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and then encoded. The specific adjustment process is: the behavior effectiveness factor of the unmanned vehicle in the first monitoring period is processed by ratio with the behavior effectiveness threshold to obtain the behavior effectiveness ratio of the unmanned vehicle in the first monitoring period, and match it with the adjustment set corresponding to each behavior effectiveness ratio interval stored in the coding database, thereby obtaining the adjustment set of the unmanned vehicle in the first monitoring period; the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted by the adjustment set, and re-encoded; in an example embodiment, it is assumed that the behavior of the unmanned vehicle in the first monitoring period is The effective ratio is P, and the behavior effective ratio interval [P-16%, P+4%] stored in the coding database, the corresponding adjustment set of the behavior effective ratio interval [P-16%, P+4%] includes: perception parameter adjustment part: use filtering algorithms such as mean filtering, median filtering and Kalman filtering to remove noise and redundant information in sensor data, use data fusion algorithms such as weight-based fusion and probability-based fusion to align and fuse data from different sensors, use machine learning and deep learning algorithms such as convolutional neural network (CNN) and target detection algorithms such as YOLOv5 to process the perceived data The adjustment part of the unmanned vehicle is to increase the gain of the camera by 5% to 10%, increase the exposure time by 10% to 20%, and increase the transmit power of the lidar by 5%. Therefore, the adjustment set of the unmanned vehicle in the first monitoring cycle includes: Perception parameter adjustment part: Use filtering algorithms such as mean filtering, median filtering, and Kalman filtering to remove noise and redundant information in sensor data, use data fusion algorithms such as weight-based fusion and probability-based fusion to align and fuse data from different sensors, and use machine learning and deep learning algorithms such as convolutional neural networks (CNN) and YOL Ov5 and other target detection algorithms are used to process and identify the perceived data. The unmanned vehicle adjustment part: increases the camera gain by 5% to 10%, increases the exposure time by 10% to 20%, and increases the laser radar's transmission power by 5%. The perception parameters of the perception layer of the unmanned vehicle are adjusted and re-encoded through the perception parameter adjustment part of the unmanned vehicle's adjustment set in the first monitoring cycle. At the same time, the unmanned vehicle is adjusted through the unmanned vehicle adjustment part of the unmanned vehicle's adjustment set in the first monitoring cycle, and the operation of the unmanned vehicle in the second monitoring cycle is continued to be monitored.

[0080] Furthermore, the operation scenario modeling of the unmanned vehicle is completed in the following specific process:

[0081] The operation process of the unmanned vehicle is obtained, divided into a number of monitoring periods according to equal time intervals, and the first monitoring period is extracted by time sequence numbering, thereby encoding the perception parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period; the above-mentioned operation process of the unmanned vehicle refers to the detailed process of the unmanned vehicle perceiving the environment, encoding behavior and modeling the scene of its operation. The specific process, duration and details of this series of processes are comprehensively formulated by the modeling engineer based on factors such as vehicle performance, road conditions and safety standards; the above-mentioned encoding of the perception parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring period refers to encoding through a coding model preset by the modeling engineer. In an exemplary embodiment In the figure, part of the coding in a coding model is as follows: the first step is to determine whether the vehicle is started or stopped, with starting being 1 and stopping being 0, which has the highest priority; the second step is to determine the vehicle's driving direction (00, 01, 10, 11 for four directions A, B, C, and D); the third step is to determine the lane topology type (eight types are 000, 001, 010, 011, 100, 101, 110, 111); the fourth step is to determine the driving behavior (four typical driving behaviors are 00, 01, 10, 11); the fifth step is to determine the vehicle's driving mode (straight ahead, right turn, left turn, and U-turn are 00, 01, 10, 11); the sixth step is to establish the current road scene model and continue to return to the first step, otherwise the road modeling is terminated.

[0082] Through the above process, the perception parameters of the perception layer of the unmanned vehicle within several monitoring cycles are encoded and integrated, and the integrated code is marked as the total code of the operation scenario of the unmanned vehicle; the operation scenario modeling of the unmanned vehicle is completed by decoding the total code of the operation scenario of the unmanned vehicle; the above decoding of the total code of the operation scenario of the unmanned vehicle refers to decoding using the decoding model corresponding to the encoding model preset by the modeling engineer, such as 1100010001, which means 1 10 001 00 01, which means starting the super C direction and keeping the right turn in the right turn lane.

[0083] The coding database is used to store traffic density boundary values, minimum vehicle distance boundary values, traffic speed discrete values, behavior reference sets corresponding to each risk impact interval, weight factors corresponding to perception effectiveness factors, weight factors corresponding to operation effectiveness factors, weight factors corresponding to risk impact, adjustment sets corresponding to each behavior effectiveness ratio interval, and behavior effectiveness thresholds.

[0084] In a specific embodiment, the present invention provides a vehicle driving behavior coding and scenario modeling system for unmanned driving. By comprehensively collecting various parameters of the environment in which the unmanned vehicle operates and accurately evaluating the risk impact of environmental factors on the operation of the unmanned vehicle, a suitable unmanned vehicle behavior reference set is intelligently matched. On this basis, the system further collects the performance parameters of the perception layer to which the unmanned vehicle belongs, the vehicle's operating parameters, and the real-time perception parameters of the perception layer. By comprehensively considering the perception layer performance, the vehicle's operating status, and the behavior reference set, the unmanned vehicle's effective behavior factors can be deeply analyzed and extracted. Based on the comparison of the unmanned vehicle's effective behavior factors with the effective behavior thresholds, the perception parameters of the unmanned vehicle's perception layer are finely adjusted and encoded, thereby achieving accurate modeling of the unmanned vehicle's operating scenarios, which not only significantly improves the accuracy of vehicle driving behavior coding and scenario modeling, but also improves the unmanned vehicle's environmental adaptability and decision-making accuracy, and enhances its safety and reliability in complex scenarios.

[0085] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A vehicle driving behavior coding and scene modeling system for unmanned driving, characterized in that: include: The operating environment assessment module is used to collect the operating environment parameters of the unmanned vehicle, assess the risk impact of the operating environment of the unmanned vehicle, and match the behavior reference set of the unmanned vehicle; The data perception analysis module is used to collect the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the perception parameters of the perception layer to which the unmanned vehicle belongs, and analyze the effective factors of the behavior of the unmanned vehicle by integrating the performance parameters of the perception layer to which the unmanned vehicle belongs, the operating parameters of the unmanned vehicle, and the behavior reference set of the unmanned vehicle; The operation behavior determination module is used to compare the behavior effectiveness factor of the unmanned vehicle with the behavior effectiveness threshold. If the behavior effectiveness factor of the unmanned vehicle is greater than the behavior effectiveness threshold, the perception parameters of the perception layer to which the unmanned vehicle belongs are behaviorally encoded. If the behavior effectiveness factor of the unmanned vehicle is less than or equal to the behavior effectiveness threshold, the perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and encoded, thereby completing the modeling of the operation scenario of the unmanned vehicle.

2. A vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 1, characterized in that: The evaluation of the risk impact of the environment in which the unmanned vehicle operates specifically refers to the risk impact of the environment in which the unmanned vehicle operates within the first monitoring cycle, and the specific evaluation process is as follows: Extracting the traffic density of the environment in which the unmanned vehicle is operating in the first monitoring period and the minimum vehicle distance between the unmanned vehicle and each adjacent vehicle in the first monitoring period from the parameters of the environment in which the unmanned vehicle is operating, sorting the minimum vehicle distances between the unmanned vehicle and each adjacent vehicle in the first monitoring period in ascending order, extracting the minimum vehicle distance ranked first, and marking it as the minimum vehicle distance of the unmanned vehicle in the first monitoring period; The average speed of each sampled vehicle in the environment where the unmanned vehicle operates in the first monitoring period is extracted from the parameters of the environment where the unmanned vehicle operates, and standard deviation processing is performed. The processing result is marked as the discrete value of the traffic speed in the environment where the unmanned vehicle operates in the first monitoring period, thereby comprehensively obtaining an assessment method for the risk impact of the environment where the unmanned vehicle operates.

3. A vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 1, characterized in that: The behavior reference set of the unmanned vehicle is matched, and the specific matching process is: The risk impact degree of the environment in which the unmanned vehicle operates in the first monitoring period is matched with the behavior reference set corresponding to each risk impact degree interval stored in the coding database, thereby obtaining the behavior reference set of the unmanned vehicle.

4. The vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 1, characterized in that: The collection of performance parameters of the perception layer of the unmanned vehicle, the operating parameters of the unmanned vehicle, and the perception parameters of the perception layer of the unmanned vehicle specifically refers to the collection of performance parameters of the perception layer of the unmanned vehicle during the first monitoring period, the operating parameters of the unmanned vehicle during the first monitoring period, and the perception parameters of the perception layer of the unmanned vehicle during the first monitoring period.

5. The vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 1, characterized in that: The analysis of the effective factors of the behavior of the unmanned vehicle specifically refers to the analysis of the effective factors of the behavior of the unmanned vehicle in the first monitoring period, and the specific analysis method is: The effective perception factor of the unmanned vehicle in the first monitoring period is analyzed by integrating the performance parameters of the perception layer of the unmanned vehicle in the first monitoring period and the behavior reference set of the unmanned vehicle; The operating parameters of the unmanned vehicle in the first monitoring period and the behavior reference set of the unmanned vehicle are integrated to analyze the effective operating factors of the unmanned vehicle in the first monitoring period; The perception effectiveness factor of the unmanned vehicle in the first monitoring period, the operation effectiveness factor of the unmanned vehicle in the first monitoring period and the risk impact degree of the environment in which the unmanned vehicle operates in the first monitoring period are comprehensively processed to obtain an analysis method for the behavior effectiveness factor of the unmanned vehicle in the first monitoring period.

6. A vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 5, characterized in that: The specific analysis method of the behavior effectiveness factor of the unmanned vehicle in the first monitoring period is as follows: ； in, is the effective factor of the behavior of the unmanned vehicle in the first monitoring cycle, is the perception effectiveness factor of the unmanned vehicle in the first monitoring cycle, is the effective operation factor of the unmanned vehicle in the first monitoring cycle, is the risk impact of the environment in which the unmanned vehicle operates in the first monitoring cycle, e is a natural constant, is the weight factor corresponding to the perception effectiveness factor preset in the coding database, is the weight factor corresponding to the running effective factor preset in the coding database, It is the weight factor corresponding to the risk impact preset in the coding database.

7. A vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 6, characterized in that: The specific analysis process of the perception effectiveness factor of the unmanned vehicle in the first monitoring cycle is as follows: Extracting the minimum perception area of ​​the perception layer of the unmanned vehicle in the first monitoring period and the average interval length of data refresh of the perception layer of the unmanned vehicle in the first monitoring period from the performance parameters of the perception layer of the unmanned vehicle in the first monitoring period; Extracting the total number of tracking frames of the perception layer of the unmanned vehicle in the first monitoring period and the number of tracking frames of each sampled reference object from the performance parameters of the perception layer of the unmanned vehicle in the first monitoring period, performing ratio processing on the number of tracking frames of each sampled reference object and the total number of tracking frames, and performing average processing on the results of the ratio processing, thereby marking the average processing result as the average ratio of the object tracking frames of the perception layer of the unmanned vehicle in the first monitoring period; The minimum perception area boundary value, the average data refresh interval boundary value and the object tracking frame reference average ratio are extracted from the unmanned vehicle's behavior reference set, thereby comprehensively obtaining an analysis method for the perception effectiveness factor of the unmanned vehicle in the first monitoring cycle.

8. The vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 6, characterized in that: The specific analysis process of the effective operation factor of the unmanned vehicle in the first monitoring period is as follows: Extracting the average duration of command responses of the unmanned vehicle in the first monitoring period and the running trajectory curve of the unmanned vehicle in the first monitoring period from the running parameters of the unmanned vehicle in the first monitoring period, obtaining the expected running trajectory curve of the unmanned vehicle in the first monitoring period, and performing a curve overlap comparison between the running trajectory curve of the unmanned vehicle in the first monitoring period and the expected running trajectory curve of the unmanned vehicle in the first monitoring period, thereby obtaining the yaw rate of the unmanned vehicle in the first monitoring period through data analysis; Extracting the direction angle of the unmanned vehicle at each detection time point from the operating parameters of the unmanned vehicle in the first monitoring cycle, extracting the expected direction angle of the unmanned vehicle at each detection time point from the behavior reference set of the unmanned vehicle, performing difference processing with the direction angle of the unmanned vehicle at the corresponding detection time point, performing absolute value processing on the difference processing, accumulating the results of the absolute value processing, and marking them as the total deviation value of the direction angle of the unmanned vehicle in the first monitoring cycle; At the same time, the average command response time limit, yaw rate limit value and azimuth total deviation limit value are extracted from the unmanned vehicle's behavior reference set, and an analysis method for the effective operating factors of the unmanned vehicle in the first monitoring cycle is obtained comprehensively.

9. The vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 1, characterized in that: The perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted and then encoded. The specific adjustment process is as follows: The behavior validity factor of the unmanned vehicle in the first monitoring period is processed by ratio with the behavior validity threshold to obtain the behavior validity ratio of the unmanned vehicle in the first monitoring period, and the adjustment set corresponding to each behavior validity ratio interval stored in the coding database is matched to obtain the adjustment set of the unmanned vehicle in the first monitoring period; The perception parameters of the perception layer to which the unmanned vehicle belongs are adjusted through the adjustment set and re-encoded.

10. The vehicle driving behavior coding and scene modeling system for unmanned driving according to claim 1, characterized in that: The specific process of completing the modeling of the operation scenario of the unmanned vehicle is as follows: The operation process of the unmanned vehicle is obtained, divided into a number of monitoring cycles according to equal time intervals, and the first monitoring cycle is extracted by time sequence numbering, thereby encoding the perception parameters of the perception layer to which the unmanned vehicle belongs in the first monitoring cycle; Through the above process, the perception parameters of the perception layer of the unmanned vehicle in several monitoring cycles are encoded and integrated, and the integrated code is marked as the total code of the operation scenario of the unmanned vehicle; The operation scenario modeling of the unmanned vehicle is completed by decoding the total code of the operation scenario of the unmanned vehicle.

Images