Automatic driving evaluation system based on artificial intelligence
Through an AI-based autonomous driving evaluation system, meteorological models and sensor parameter adjustments are used to simulate extreme weather scenarios for quantitative evaluation of autonomous driving systems, which addresses the shortcomings of traditional evaluation methods and achieves more accurate performance evaluation and optimization recommendations.
Patent Information
- Application Number
- CN202510698330.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-28
- Publication Date
- 2025-09-16
- Estimated Expiration
- 2045-05-28
AI Technical Summary
Traditional autonomous driving system evaluation methods have shortcomings in extreme weather conditions, making it difficult to comprehensively and accurately evaluate performance. They ignore the accuracy of sensor data and the rationality of the system's decision-making in extreme weather conditions, resulting in evaluation results that are not comprehensive and objective.
An AI-based autonomous driving evaluation system is used to obtain meteorological data through the meteorological model coupling module. The sensor parameter adjustment module automatically adjusts sensor performance. The autonomous driving simulation module operates in extreme weather scenarios. The evaluation and analysis module uses artificial intelligence algorithms to perform quantitative evaluation and generate detailed evaluation reports.
It significantly improves the accuracy of performance evaluation of autonomous driving systems under extreme weather conditions, provides a comprehensive and dynamic evaluation method, generates detailed evaluation reports, points out the strengths and weaknesses of the system, and provides a reference for optimization.
Smart Images

Figure CN120653945A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of autonomous driving technology, and in particular to an autonomous driving evaluation system based on artificial intelligence. Background Art
[0002] With the rapid development of autonomous driving technology, its safety and reliability have become the focus of public and industry attention. Autonomous driving systems need to operate stably under various complex and extreme weather conditions to ensure the safety of passengers and pedestrians. However, extreme weather conditions such as heavy rain, dense fog, and heavy snow can interfere with vehicle sensors, affecting their ability to perceive the environment, and thus affecting the decision-making and control accuracy of the autonomous driving system.
[0003] Traditional autonomous driving system evaluation methods often rely on laboratory simulations or limited field tests. These methods have obvious shortcomings when simulating extreme weather conditions. Laboratory simulations cannot fully reproduce the complexity and dynamics of the real world, while field tests are affected by many uncontrollable factors such as weather and traffic conditions. It is difficult to comprehensively and accurately evaluate the performance of autonomous driving systems under extreme weather conditions. In addition, traditional evaluation methods often focus on single-dimensional performance indicators, such as driving safety or decision rationality, while ignoring other important factors such as sensor data accuracy, resulting in evaluation results that are not comprehensive and objective.
[0004] Aiming at the shortcomings of traditional autonomous driving system evaluation methods under extreme weather conditions, the present invention proposes an autonomous driving evaluation system based on artificial intelligence. Summary of the Invention
[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide an artificial intelligence-based autonomous driving evaluation system. It can establish a data connection with the external meteorological model through the meteorological model coupling module, obtain comprehensive meteorological data in real time, and construct simulated extreme weather scenarios based on these data. At the same time, the sensor parameter adjustment module automatically adjusts the performance parameters of the vehicle sensor according to different weather types to ensure the accuracy of the sensor data under extreme weather conditions. The autonomous driving simulation module runs the autonomous driving system model in the simulated extreme weather scenario to simulate the vehicle driving process. Finally, the evaluation and analysis module uses artificial intelligence algorithms to analyze the operating data and quantitatively evaluate the performance of the autonomous driving system in the simulated extreme weather scenarios, thereby providing a comprehensive and dynamic evaluation method, which significantly improves the accuracy of the performance evaluation of the autonomous driving system under extreme weather conditions.
[0006] To solve the above technical problems, the present invention provides the following technical solutions: an artificial intelligence-based autonomous driving evaluation system, which includes the following components: a weather model coupling module, a sensor parameter adjustment module, an autonomous driving simulation module, and an evaluation and analysis module;
[0007] The meteorological model coupling module is used to establish a data connection with an external meteorological model to obtain meteorological data in real time, including but not limited to temperature, humidity, wind speed, rainfall, snowfall, and visibility information, and to construct simulated extreme weather scenarios based on this meteorological data;
[0008] The sensor parameter adjustment module: When the meteorological model coupling module generates extreme weather scenarios, this module automatically adjusts the performance parameters of the vehicle sensors according to different weather types;
[0009] The autonomous driving simulation module runs the autonomous driving system model in a simulated extreme weather scenario based on the adjusted sensor performance parameters, simulating the vehicle's driving process in that environment, including the vehicle's acceleration, deceleration, steering, and avoidance maneuvers, as well as the autonomous driving system's perception, decision-making, and control of environmental information.
[0010] The evaluation and analysis module uses artificial intelligence algorithms to analyze the operating data of the autonomous driving simulation module, and quantitatively evaluates the performance of the autonomous driving system in simulated extreme weather scenarios based on set evaluation indicators such as vehicle driving safety, decision-making rationality, and sensor data accuracy. It generates a detailed evaluation report that points out the advantages and disadvantages of the autonomous driving system in responding to climate change.
[0011] Furthermore, when the meteorological model coupling module constructs a simulated extreme weather scenario, a dynamic climate parameter generation algorithm is adopted. The algorithm is based on the fusion analysis of historical meteorological data and real-time meteorological data, and generates the simulated extreme weather scenario parameter S through the formula S=α×H+(1-α)×R, where H is the parameter set corresponding to the extreme weather type in the historical meteorological data, R is the parameter set of real-time meteorological data, and α is the historical data weight coefficient with a value range of [0.3, 0.7]. The determination method is: statistically analyze the goodness of fit of historical data and real-time data when extreme weather occurred in the target area in the past ten years, and use the optimization algorithm to find the α value with the highest goodness of fit, so that the generated simulated extreme weather scenario is more in line with the actual situation, thereby improving the authenticity and reliability of subsequent evaluation.
[0012] Furthermore, the sensor parameter adjustment module uses a dynamic adjustment algorithm based on visual fuzziness when adjusting the camera imaging quality parameters. For a rainstorm scene, the camera image clarity adjustment formula is:
[0013]
[0014] Among them C old is the camera image clarity before adjustment, C newis the adjusted clarity, β is the rainfall influence coefficient, and its value range is [0.1, 0.5]. The coefficient is determined by collecting the actual imaging quality data of the camera under different rainfall intensities and using regression analysis. V is the real-time rainfall, V max For the set maximum rainfall threshold, the algorithm can dynamically adjust the camera clarity according to the real-time rainfall, more accurately simulate the interference of heavy rain on camera imaging, and provide more realistic sensor data input for autonomous driving simulation.
[0015] Furthermore, the algorithm for evaluating vehicle driving safety in the evaluation and analysis module adopts a multi-dimensional risk assessment model, and its calculation formula is:
[0016] R s =ω1×D+ω2×A+ω3×L
[0017] where R s is the vehicle driving safety risk value, D is the real-time distance to the preceding vehicle or obstacle, A is the vehicle acceleration, L is the lane deviation degree, ω1, ω2, and ω3 are the weights of the corresponding parameters, and the value ranges are [0.4, 0.6], [0.2, 0.4], and [0.2, 0.4], respectively. The weights are determined by collecting a large amount of actual traffic accident case data, analyzing the influence of various factors on the occurrence of accidents, and using the hierarchical analysis method to determine the weights of each factor. This model can comprehensively and accurately evaluate the driving safety risks of vehicles in simulated extreme weather scenarios, providing a scientific basis for the safety evaluation of autonomous driving systems.
[0018] Furthermore, when evaluating the rationality of the autonomous driving system's decision, the evaluation and analysis module adopts a decision evaluation algorithm based on the semantic understanding of traffic rules. First, the traffic rules are semantically parsed and a rule knowledge base is constructed. For each decision of the autonomous driving system, the algorithm is used to evaluate the rationality of the autonomous driving system's decision. Calculate the probability P that the decision complies with traffic rules r , where N is the total number of traffic rules involved in the decision, M i is the matching degree of the i-th rule. When the decision fully complies with the rule, M i =1, if it is partially consistent, a value between [0, 1) is assigned according to the degree of consistency, and if it is completely inconsistent, M i =0, by combining expert experience with actual traffic data, the judgment criteria for the matching degree of each rule are determined, thereby achieving a quantitative evaluation of the rationality of the autonomous driving system's decision-making and effectively evaluating whether the system's decision-making is compliant in extreme weather conditions.
[0019] Furthermore, the evaluation and analysis module uses the interactive experience comfort evaluation index when evaluating the passenger comfort of the autonomous driving system:
[0020] Decision transparency: The autonomous driving system should inform passengers of its decision logic in real time through the instrument panel or voice, with a coverage rate of ≥90%;
[0021] Path planning rationality: Avoid frequent lane changes or choosing bumpy roads. Generate a comfortable path model through training with historical navigation data and calculate the deviation between the actual path and the comfortable path.
[0022] Emergency response softness: When avoiding obstacles or braking, the vehicle uses a pre-aiming algorithm to adjust the operating force in advance to avoid sudden braking or sudden steering. This can be quantitatively evaluated using passenger body displacement sensor data.
[0023] Furthermore, the autonomous driving simulation module adopts a dynamic environment adaptation algorithm based on reinforcement learning during the simulation of vehicle driving, defines the state space as the vehicle's current position, speed, sensor data and surrounding environment information, and the action space as the vehicle's acceleration, deceleration, and steering operations. Update the action-value function Q(s,a), where η is the learning rate, ranging from 0.1 to 0.3, and γ is the discount factor, ranging from 0.7 to 0.9. These two parameters are optimized and determined by testing in a large number of simulated scenarios, with the goal of enabling the system to reach stable and reasonable decisions as quickly as possible. r is the immediate reward obtained for the current action, and the reward mechanism is set based on whether a collision occurs or whether the vehicle deviates from the lane. This algorithm enables the autonomous driving simulation module to continuously learn and adapt to dynamically changing environments in simulated extreme weather scenarios, improving the authenticity and reliability of the simulation and providing richer and more effective data for the evaluation and analysis module.
[0024] Furthermore, the sensor parameter adjustment module uses a penetration adjustment algorithm based on particle scattering theory when adjusting the performance parameters of the laser radar. In a dense fog scene, the laser radar penetration rate P t The calculation formula is P t =e -σ×N×d , where σ is the scattering cross section of the laser by fog droplets. This parameter is obtained by data fitting through laser scattering experiments on fog droplets of different concentrations. N is the number of fog droplets per unit volume, which can be determined based on the dense fog scene parameters generated by the meteorological model coupling module. d is the laser propagation distance. This algorithm is based on physical principles and can accurately simulate the weakening effect of dense fog on LiDAR signals. This makes the sensor parameter adjustment more consistent with actual physical laws, improving the accuracy of the LiDAR performance evaluation system of the autonomous driving evaluation system.
[0025] Furthermore, when generating an evaluation report, the evaluation analysis module adopts a report generation algorithm based on natural language generation, which specifically includes: first, performing semantic conversion processing on the quantitative evaluation results, constructing a mapping relationship between different evaluation dimensions and semantic descriptions, and then determining a weight distribution mechanism based on the importance of each evaluation dimension. This mechanism scores the degree of influence of each evaluation dimension on the overall performance of the autonomous driving system by industry experts, and uses the Delphi method to perform statistical analysis on the scoring results. The results of each evaluation dimension are weighted and fused according to the determined weights to generate a thematic framework of the evaluation text. Finally, a pre-trained language model is used to generate a complete evaluation report text containing targeted improvement suggestions based on the generated thematic framework and in combination with specific simulation data cases.
[0026] Furthermore, after obtaining meteorological data, the meteorological model coupling module also performs spatiotemporal interpolation processing on the data, using an optimization algorithm based on Kriging interpolation, specifically including: for locations in the target area where no meteorological monitoring points are set, constructing a spatial interpolation model based on the data of known monitoring points, and determining the influence weight of each known monitoring point on the estimated value of the meteorological parameter at the target location by analyzing the spatial autocorrelation of meteorological parameters. The weight determination process is implemented by minimizing the objective function of the estimated variance, and the Lagrange multiplier method is used to solve the objective function. The meteorological parameter values of each known monitoring point are weightedly combined and calculated using the determined weights to obtain the estimated value of the meteorological parameter at the target location. The interpolation results are cross-validated, and the interpolation model is optimized and adjusted by comparing the estimation errors of different interpolation methods at the verification points, thereby improving the spatial integrity and accuracy of the meteorological data, making the constructed simulated extreme weather scenes more uniform and more realistic in the region, and providing a better data foundation for subsequent autonomous driving evaluation.
[0027] Compared with existing technologies, this AI-based autonomous driving evaluation system has the following beneficial effects:
[0028] 1. The system establishes a data connection with the external meteorological model through the meteorological model coupling module, obtains comprehensive meteorological data including temperature, humidity, wind speed, rainfall, snowfall, visibility, etc. in real time, and constructs simulated extreme weather scenarios based on this data. The sensor parameter adjustment module automatically adjusts the performance parameters of the vehicle sensors according to different weather types to ensure the accuracy of sensor data under extreme weather conditions. The autonomous driving simulation module runs the autonomous driving system model in the simulated extreme weather scenario to simulate the vehicle driving process. Finally, the evaluation and analysis module uses artificial intelligence algorithms to analyze the operating data and quantitatively evaluate the performance of the autonomous driving system in simulated extreme weather scenarios. This comprehensive and dynamic evaluation method significantly improves the accuracy of the performance evaluation of the autonomous driving system under extreme weather conditions, providing strong support for the further optimization of autonomous driving technology.
[0029] Second, when generating an evaluation report through the evaluation analysis module, the system adopts a report generation algorithm based on natural language generation, performs semantic conversion on the quantitative evaluation results, and constructs a mapping relationship between different evaluation dimensions and semantic descriptions. By determining a weight distribution mechanism based on the importance of each evaluation dimension, the results of each evaluation dimension are weighted and fused to generate a thematic framework for the evaluation text. Finally, using the pre-trained language model, based on the generated thematic framework and combined with specific simulation data cases, a complete evaluation report text containing targeted improvement suggestions is generated. This detailed evaluation report not only points out the advantages and disadvantages of the autonomous driving system in responding to climate change, but also provides specific improvement directions and suggestions, providing an important reference for the continuous optimization and upgrading of the autonomous driving system.
[0030] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.
[0032] Figure 1 Develop a flow chart for the AI-based autonomous driving evaluation system functionality;
[0033] Figure 2 Schematic diagram of the structure of the autonomous driving evaluation system based on artificial intelligence. DETAILED DESCRIPTION
[0034] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.
[0035] Example 1
[0036] An autonomous driving technology company needs to test the vehicle's obstacle avoidance capabilities in heavy rain, verify the reliability of sensors under precipitation interference, and verify the rationality of autonomous driving decisions.
[0037] Access to real-time meteorological data: If rainfall in a certain area reaches 50 mm / h (exceeding the rainstorm threshold), combined with historical rainstorm data (such as the extreme rainfall parameter set H for the same season in previous years), a dynamic climate parameter generation algorithm is used to construct a simulated rainstorm scenario:
[0038] S=α×H+(1-α)×R
[0039] Among them, the historical data weight coefficient α=0.4, and the real-time data R includes parameters such as rainfall and visibility.
[0040] After identifying a rainstorm scene, the camera image quality dynamic adjustment algorithm is triggered: Assume that the clarity before adjustment is C old =80%, rainfall influence coefficient β = 0.3, real-time rainfall V = 50 mm / h, maximum threshold V max =60mm / h, then the clarity after adjustment
[0041] In a simulated vehicle driving in a rainstorm scenario, a stationary obstacle suddenly appears in front of the vehicle. The system adjusts the steering action using a reinforcement learning algorithm. The state space includes the vehicle speed (20m / s) and the obstacle distance detected by the sensor (30 meters). The action space selects steering avoidance, and the action value function is updated using the Q-learning formula: If the obstacle is successfully avoided and the lane is not deviated, the immediate reward r = +10, the learning rate η = 0.1, and the discount factor γ = 0.9.
[0042] Calculate the safety risk value using a multi-dimensional risk assessment model: R s =ω1×D+ω2×A+ω3×L Assume weights ω1=0.5 (distance), ω2=0.3 (acceleration), ω3=0.2 (lane departure), measured distance D=10 meters, acceleration A=-5m / s 2 (deceleration), deviation degree L = 0.2 meters, calculate R s=0.5×10+0.3×(-5)+0.2×0.2=3.54 (the lower the value, the safer).
[0043] Based on the traffic rule semantic understanding algorithm, determine whether the steering avoidance decision complies with the rule of "reducing speed and maintaining distance in heavy rain" assuming that two rules are involved, with matching degrees of 0.8 and 0.9 respectively, then the probability of compliance is:
[0044] Example 2
[0045] A car company conducted autonomous driving tests in a typical foggy area in winter. The goal was to verify the vehicle's ability to detect long-distance obstacles, sensor data reliability, and system emergency response mechanism in an extremely dense fog environment with visibility below 50 meters. The test scenario was set at 6 o'clock in the morning (poor lighting conditions), the temperature was 5°C, the air humidity reached 95%, and thick fog continued to cover the test section, simulating the challenges of real winter smog weather to autonomous driving.
[0046] Access the data from the regional meteorological monitoring network to identify the current environmental parameters, including visibility of 40 meters, average droplet size of 10 microns, and high droplet concentration per unit volume.
[0047] Combined with the historical data of winter dense fog in the area in the past five years (such as fog duration and spatial distribution characteristics), the meteorological parameters of the uncovered monitoring points in the test section were supplemented through the spatiotemporal interpolation algorithm (based on Kriging interpolation optimization), and a high-precision dense fog distribution simulation scene was generated to clarify the differences in droplet density at different locations.
[0048] Send an "extremely dense fog weather" trigger signal to the system to start the sensor adaptive adjustment process.
[0049] After receiving the dense fog scene signal, it automatically identifies the type of sensor that needs to be adjusted: LiDAR (most affected by fog droplet scattering).
[0050] Dynamically increase the laser emission power to the upper limit of safety to enhance signal strength to combat fog droplet attenuation;
[0051] Adjust the lidar scanning mode from "wide-area scanning" to "focused scanning", prioritize enhancing detection accuracy within 200 meters ahead, enable the built-in anti-interference filtering algorithm, filter out noise data generated by fog droplet scattering, improve point cloud purity, and simultaneously send sensor status update information (such as the adjusted detection distance threshold and data refresh rate) to the autonomous driving simulation module.
[0052] After loading the dense fog scene, the simulated vehicle travels on two lanes in both directions at a speed of 40 km / h (lower than the regular speed limit and in compliance with low visibility safety requirements).
[0053] The system builds an environmental model in real time through lidar point cloud data. When it detects a fuzzy point cloud (suspected to be the rear of a truck) 150 meters ahead, it triggers the multi-sensor fusion verification mechanism. The millimeter-wave radar simultaneously scans the area to confirm the presence of metal reflection signals. The camera activates the super-resolution algorithm and attempts to capture the outline of the object through the gaps in the fog curtain (due to limited visibility, only a fuzzy image is obtained). Based on the fused data, the autonomous driving system determines it as a "stationary obstacle" and triggers an emergency braking decision. The motor controller outputs maximum braking force, the tire anti-lock braking system (ABS) is activated simultaneously, and the vehicle posture control system adjusts the suspension damping to suppress the body pitch during emergency braking and maintain driving stability.
[0054] Comparing the obstacle distance output by the lidar (145 meters) with the actual position (150 meters), the calculated error rate was 3.3%, which was judged to be within an acceptable range in a dense fog environment. Analysis of the integrity of the point cloud data showed that due to the decrease in penetration, the top and edge point clouds of the obstacle were missing, which required subsequent algorithm interpolation optimization. The vehicle took 4.2 seconds to completely stop from 40 km / h, and the braking distance was 38 meters, which was less than the theoretical safe distance (45 meters, calculated based on the road friction coefficient). The braking performance was determined to be up to standard, and it was confirmed that the system completed the "perception-decision-execution" link within 0.2 seconds after detecting the obstacle, which met the "emergency response time <0.5 seconds" requirement in the autonomous driving safety standard.
[0055] LiDAR has basic obstacle detection capabilities in extremely dense fog environments, but the accuracy of long-distance point clouds decreases significantly, and it needs to be combined with millimeter-wave radar to improve reliability. The emergency braking strategy is reasonable, but it is recommended to optimize the multi-sensor fusion algorithm to reduce the risk of misjudgment in low visibility.
[0056] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.
Claims
1. An AI-based autonomous driving evaluation system, characterized by: The system includes the following components: a weather model coupling module, a sensor parameter adjustment module, an autonomous driving simulation module, and an evaluation and analysis module; The meteorological model coupling module is used to establish a data connection with an external meteorological model to obtain meteorological data in real time, including but not limited to temperature, humidity, wind speed, rainfall, snowfall, and visibility information, and to construct simulated extreme weather scenarios based on this meteorological data; The sensor parameter adjustment module: When the meteorological model coupling module generates extreme weather scenarios, this module automatically adjusts the performance parameters of the vehicle sensors according to different weather types; The autonomous driving simulation module runs the autonomous driving system model in a simulated extreme weather scenario based on the adjusted sensor performance parameters, simulating the vehicle's driving process in that environment, including the vehicle's acceleration, deceleration, steering, and avoidance maneuvers, as well as the autonomous driving system's perception, decision-making, and control of environmental information. The evaluation and analysis module uses artificial intelligence algorithms to analyze the operating data of the autonomous driving simulation module, and quantitatively evaluates the performance of the autonomous driving system in simulated extreme weather scenarios based on set evaluation indicators, such as vehicle driving safety, decision-making rationality, sensor data accuracy, and passenger comfort. It generates a detailed evaluation report that points out the advantages and disadvantages of the autonomous driving system in responding to climate change.
2. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: When the meteorological model coupling module constructs a simulated extreme weather scenario, a dynamic climate parameter generation algorithm is adopted. The algorithm is based on the fusion analysis of historical meteorological data and real-time meteorological data, and generates the simulated extreme weather scenario parameter S through the formula S = α × H + (1-α) × R, where H is the parameter set corresponding to the extreme weather type in the historical meteorological data, R is the parameter set of real-time meteorological data, and α is the weight coefficient of historical data.
3. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: The sensor parameter adjustment module uses a dynamic adjustment algorithm based on visual fuzziness when adjusting the camera imaging quality parameters. For rainstorm scenes, the camera image clarity adjustment formula is: Among them C old is the camera image clarity before adjustment, C new is the adjusted clarity, β is the rainfall influence coefficient, V is the real-time rainfall, V max is the maximum rainfall threshold set.
4. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: The algorithm used in the evaluation and analysis module to evaluate vehicle driving safety adopts a multi-dimensional risk assessment model, and its calculation formula is: R s =ω1×D+ω2×A+ω3×L where R s is the vehicle driving safety risk value, D is the real-time distance to the preceding vehicle or obstacle, A is the vehicle acceleration, L is the lane deviation degree, and ω1, ω2, and ω3 are the weights of the corresponding parameters.
5. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: When evaluating the rationality of the autonomous driving system's decision, the evaluation and analysis module adopts a decision evaluation algorithm based on the semantic understanding of traffic rules. First, the traffic rules are semantically parsed and a rule knowledge base is constructed. For each decision of the autonomous driving system, the formula Calculate the probability P that the decision complies with traffic rules r , where N is the total number of traffic rules involved in the decision, M i is the matching degree of the i-th rule.
6. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: The evaluation and analysis module uses the interactive experience comfort evaluation index when evaluating the passenger comfort of the autonomous driving system: Decision transparency: The autonomous driving system should inform passengers of its decision logic in real time through the instrument panel or voice, with a coverage rate of ≥90%; Path planning rationality: Avoid frequent lane changes or choosing bumpy roads. Generate a comfortable path model through training with historical navigation data and calculate the deviation between the actual path and the comfortable path. Emergency response softness: When avoiding obstacles or braking, the vehicle uses a pre-aiming algorithm to adjust the operating force in advance to avoid sudden braking or sudden steering. This can be quantitatively evaluated using passenger body displacement sensor data.
7. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: The autonomous driving simulation module adopts a dynamic environment adaptation algorithm based on reinforcement learning in the process of simulating vehicle driving. The state space is defined as the vehicle's current position, speed, sensor data and surrounding environment information, and the action space is the vehicle's acceleration, deceleration and steering operations. Update the action value function Q(s,a), where η is the learning rate, γ is the discount factor, and r is the immediate reward obtained for the current action. The reward mechanism is set based on whether a collision occurs or whether the vehicle deviates from the lane.
8. The automatic driving evaluation system based on artificial intelligence according to claim 1, characterized in that: The sensor parameter adjustment module uses a penetration adjustment algorithm based on particle scattering theory when adjusting the performance parameters of the laser radar. In dense fog scenes, the laser radar penetration rate P t The calculation formula is P t =e -σ×N×d , where σ is the scattering cross section of the droplets to the laser, N is the number of droplets per unit volume, and d is the laser propagation distance.
9. The artificial intelligence-based autonomous driving evaluation system according to claim 1, characterized in that: When generating an evaluation report, the evaluation analysis module adopts a report generation algorithm based on natural language generation, which specifically includes: first, performing semantic conversion processing on the quantitative evaluation results, constructing a mapping relationship between different evaluation dimensions and semantic descriptions, then determining a weight distribution mechanism based on the importance of each evaluation dimension, performing weighted fusion processing on the results of each evaluation dimension according to the determined weight, generating a theme framework of the evaluation text, and finally using a pre-trained language model, based on the generated theme framework and combined with specific simulation data cases, generating a complete evaluation report text containing targeted improvement suggestions.
10. The automatic driving evaluation system based on artificial intelligence according to claim 1, characterized in that: After acquiring meteorological data, the meteorological model coupling module also performs spatiotemporal interpolation processing on the data using an optimization algorithm based on Kriging interpolation. Specifically, the algorithm includes: constructing a spatial interpolation model based on data of known monitoring points for locations within the target area where no meteorological monitoring points are set; determining the influence weight of each known monitoring point on the estimated value of the meteorological parameter at the target location by analyzing the spatial autocorrelation of meteorological parameters; performing a weighted combination calculation on the meteorological parameter values of each known monitoring point using the determined weight to obtain the estimated value of the meteorological parameter at the target location; performing cross-validation processing on the interpolation results; and optimizing and adjusting the interpolation model by comparing the estimation errors of different interpolation methods at the validation points.
Citation Information
Patent Citations
Method for adjusting dynamic exposure of camera in automatic driving
CN105554410A
Man-machine perception and decision-making method based on generalized driving behavior map
CN118839588A
Performance testing device and method for intelligent networked automobile
CN118882679A
Automatic driving test system and method for special vehicle
CN119246086A
Software and hardware decoupling analogue simulation test system
CN119669066A
Cited By
Intelligent driving perception data processing method and device, electronic equipment and vehicle
CN121071438A
Forward collision early warning processing equipment and method for intelligent network connection automobile
CN121291477A
Sensor performance parameter configuration method based on intelligent driving scene
CN121327604A