A perception fusion arbitration method and system based on vehicle-to-everything (V2X) networks

Abstract

This invention provides a perception fusion arbitration method and system based on vehicle-to-everything (V2X) communication. The method includes: acquiring a ground truth model of the perception system for the road environment and meteorological information; acquiring perception information and data quality parameters from each sensor in the vehicle-mounted perception system; retrieving the influence coefficients of each sensor corresponding to the meteorological information and calculating the perception quality score of each sensor based on the data quality parameters to determine the optimal sensor; selecting candidate arbitration ratios from a pre-set perception fusion arbitration ratio case library; performing simulated fusion calculations on the perception information; comparing the matching degree of each fusion result with the ground truth model of the perception system to determine the optimal arbitration ratio; and sending the result to the vehicle controller. This invention achieves adaptive output of the optimal vehicle-mounted sensor fusion weight ratio according to the specific meteorological environment, significantly reducing the perception misidentification rate in adverse environments such as rain, snow, fog, and sandstorms, ensuring the perception prediction accuracy of the autonomous driving system in all weather conditions.

Description

Technical Field

[0001] This invention belongs to the field of autonomous driving technology, and in particular relates to a perception fusion arbitration method and system based on vehicle networking. Background Technology

[0002] With the continuous development of autonomous driving technology, autonomous vehicles are gradually being applied. In order to perceive the surrounding road environment, autonomous vehicles are usually equipped with various types of onboard sensors, such as cameras, millimeter-wave radar, and lidar, and use perception fusion technology to fuse the perception information from multiple sensors to obtain more comprehensive and accurate environmental perception results.

[0003] However, most existing autonomous driving perception fusion solutions use fixed fusion models or fixed weight allocation methods for calculation, failing to fully consider the differentiated impact of external weather conditions on the performance of various onboard sensors. For example, in rain, snow, and fog, the image quality of onboard cameras deteriorates significantly, while millimeter-wave radar is relatively less affected by weather. In dusty environments, lidar point cloud data may be interfered with by particulate matter, and the contrast of images acquired by cameras is also reduced. Applying the same perception fusion model to different road and weather conditions will cause the fusion output perception results to deviate from the actual environmental state, easily leading to inaccurate target detection or even misidentification, thus posing safety hazards to the decision-making and control of autonomous vehicles. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention aims to propose a perception fusion arbitration method and system based on vehicle networking, which solves the problem that the perception fusion model of autonomous vehicles in the prior art cannot adaptively follow changes in the weather environment, resulting in low accuracy of fused perception and affecting driving safety.

[0005] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0006] A perception fusion arbitration method based on vehicle-to-everything (V2X) communication includes: Acquire the true value model of the road environment perception system and meteorological information; Acquire the sensing information and data quality parameters of each sensor in the vehicle sensing system; The influence coefficients of each sensor corresponding to the meteorological information are retrieved, and the perception quality score of each sensor is calculated in combination with the data quality parameters. Based on the perception quality score, the optimal sensor for the current road environment is determined. Based on the optimal sensor, candidate arbitration ratios are selected from a preset case library of perception fusion arbitration ratios; each candidate arbitration ratio is used to simulate and fuse the perception information, and the matching degree of each fusion result is compared with the true value model of the perception system to determine the optimal arbitration ratio. The optimal arbitration ratio is sent to the vehicle controller for fusion calculation of its onboard perception system.

[0007] Furthermore, the ground truth model for the perception system for acquiring the road environment specifically includes: Acquire 2D perception information from roadside cameras and 3D point cloud information from roadside lidar; Visual perception fusion is performed on the 2D perception information and 3D point cloud information to construct a truth model of the perception system.

[0008] Furthermore, the meteorological information is collected by roadside meteorological sensors, and the meteorological information includes at least one of rain, snow, fog, dust, and clear weather.

[0009] Furthermore, the data quality parameters include an image sharpness coefficient; the perceived quality score includes an image quality score. The calculation process for the image quality score is as follows: The image quality score for each sensor is calculated by multiplying the influence coefficient of each sensor under the current meteorological information with the image sharpness coefficient of the corresponding sensor.

[0010] Furthermore, the sensors in the vehicle perception system include vehicle cameras, vehicle millimeter-wave radar, and vehicle lidar.

[0011] Furthermore, the optimal sensor for the current road environment is determined based on the perceived quality score, specifically including: The sensor with the highest perception quality score is determined as the optimal sensor by comparing the perception quality scores of each sensor. When multiple identical highest scores exist, the optimal sensor is determined according to the priority order of vehicle-mounted camera, vehicle-mounted LiDAR, and vehicle-mounted millimeter-wave radar.

[0012] Furthermore, candidate arbitration ratios are selected from a pre-defined perceptual fusion arbitration ratio case library, specifically including: The perception fusion arbitration comparison case library contains multiple arbitration comparison cases, and each arbitration comparison case defines the weight ratio of the vehicle camera, vehicle lidar and vehicle millimeter-wave radar in the perception system fusion calculation; From the perception fusion arbitration ratio case library, a preset number of cases with the highest corresponding optimal sensor weight ratio are selected as the candidate arbitration ratio.

[0013] A perception fusion arbitration system based on vehicle-to-everything (V2X) communication includes: The truth model acquisition module is used to acquire the truth model of the road environment perception system and meteorological information; The vehicle controller is used to acquire the sensing information and data quality parameters of each sensor in the vehicle sensing system. The cloud platform is used to retrieve the influence coefficients of each sensor corresponding to the meteorological information, and calculate the perception quality score of each sensor in combination with the data quality parameters, and determine the optimal sensor under the current road environment based on the perception quality score. Based on the optimal sensor, candidate arbitration ratios are selected from a preset case library of perception fusion arbitration ratios; each candidate arbitration ratio is used to simulate and fuse the perception information, and the matching degree of each fusion result is compared with the true value model of the perception system to determine the optimal arbitration ratio. The distribution module distributes the optimal arbitration ratio to the vehicle controller for fusion calculation of its on-board perception system.

[0014] Furthermore, the truth model acquisition module includes: Roadside cameras are installed above roadside poles to collect road environment information; Roadside lidar is deployed above roadside poles to collect road environment information; Edge computing units are used to receive road environment information collected by roadside cameras and roadside lidar, and to build a true value model of the perception system; Roadside weather sensors are used to collect weather information.

[0015] Furthermore, the perception fusion arbitration system also includes a roadside communication unit and an in-vehicle communication unit, and the vehicle controller sends the perception information and data quality parameters to the in-vehicle communication unit; The vehicle-mounted communication unit sends the perceived information and data quality parameters to the cloud platform through the roadside communication unit.

[0016] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a perception fusion arbitration method based on the Internet of Vehicles.

[0017] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements a perception fusion arbitration method based on vehicle-to-everything (V2X) networks.

[0018] Compared with existing technologies, the perception fusion arbitration method and system based on vehicle networking described in this invention have the following advantages: This invention retrieves the influence coefficients of each sensor corresponding to meteorological information and calculates the perception quality score of each sensor by combining the data quality parameters obtained from the vehicle to determine the optimal sensor. Then, it selects candidate arbitration ratios from the case library and uses the true value model of the perception system to compare the matching degree to determine the optimal arbitration ratio and send it to the vehicle. By using vehicle-to-everything (V2X) technology, it achieves a deep integration of roadside objective true values, environmental meteorology, and subjective perception quality from the vehicle, overcoming the shortcomings of traditional single fixed perception fusion models that cannot adapt to environmental changes. It achieves adaptive output of the optimal vehicle sensor fusion weight ratio according to the specific meteorological environment, significantly reducing the perception misidentification rate in harsh environments such as rain, snow, fog, and sandstorms, ensuring the perception prediction accuracy of the autonomous driving system in all-weather environments, and effectively improving the driving safety of autonomous vehicles.

[0019] This invention uses the product of the influence coefficient of each sensor under the current meteorological information and the image sharpness coefficient of the corresponding sensor as the perception quality score of each sensor. This calculation step comprehensively considers macro-environmental factors (meteorological influence system) and micro-operational status (image sharpness parameter). The quantitatively calculated score can most realistically reflect the actual availability of each sensor under the current specific road conditions, enhancing the objectivity and scientific nature of the optimal sensor evaluation.

[0020] This invention selects a preset number of cases with the highest weight ratio of the corresponding optimal sensor from the case library as candidate arbitration ratios for simulation comparison, eliminating the cumbersome process of exhaustively comparing all possible ratios in the case library. It performs preliminary targeted data dimensionality reduction through the optimal sensor, extracts only a small number of high-potential high-quality cases for matching degree calculation that consumes a lot of computing power, greatly saves the computing resources of the cloud platform, significantly reduces the processing latency of the arbitration logic, and perfectly meets the stringent requirements of real-time command issuance for autonomous driving. Attached Figure Description

[0021] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is an overall flowchart provided for embodiments of the present invention; Figure 2 This is an overall system control block diagram provided for an embodiment of the present invention; Figure 3 This is a schematic diagram of the system structure provided in an embodiment of the present invention.

[0022] Explanation of reference numerals in the attached figures: 1. Roadside camera; 2. Roadside LiDAR; 3. Edge computing unit; 4. Cloud platform; 5. Roadside RSU; 6. Weather sensor; 7. On-board unit (OBU); 8. Vehicle controller; 9. On-board camera; 10. On-board millimeter-wave radar; 11. On-board LiDAR; 12. Autonomous vehicle. Detailed Implementation

[0023] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0024] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are used only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0025] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0026] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0027] like Figures 1 to 3 As shown, a perception fusion arbitration method based on vehicle-to-everything (V2X) includes: Acquire the true value model of the road environment perception system and meteorological information; Acquire the sensing information and data quality parameters of each sensor in the vehicle sensing system; The influence coefficients of each sensor corresponding to the meteorological information are retrieved, and the perception quality score of each sensor is calculated in combination with the data quality parameters. Based on the perception quality score, the optimal sensor for the current road environment is determined. Based on the optimal sensor, candidate arbitration ratios are selected from a preset case library of perception fusion arbitration ratios; each candidate arbitration ratio is used to simulate and fuse the perception information, and the matching degree of each fusion result is compared with the true value model of the perception system to determine the optimal arbitration ratio. The optimal arbitration ratio is sent to the vehicle controller for fusion calculation of its onboard perception system.

[0028] This invention provides a perception fusion arbitration method based on vehicle-to-everything (V2X) communication. By combining current meteorological information to quantitatively evaluate the actual performance of vehicle-side sensors, the arbitration weights of perception fusion are dynamically adjusted. This enables the vehicle-mounted perception system to fuse sensors according to the most suitable sensor weights under different meteorological conditions, significantly improving the consistency between the fused perception results and the real environment, effectively avoiding misidentification, and enhancing the driving safety of autonomous vehicles in complex meteorological environments.

[0029] In the specific implementation process, the following steps should be followed: S1. Obtain the true value model of the road environment perception system and meteorological information.

[0030] The true value model of the perception system serves as the benchmark for subsequent matching degree comparison. Its construction method is as follows: receiving 2D perception information from roadside cameras and 3D point cloud information from roadside lidar; performing visual perception fusion on the 2D perception information and 3D point cloud information to construct the true value model of the perception system.

[0031] The roadside camera is mounted on top of the roadside poles. It is a 50-megapixel three-lens ultra-high-definition camera with a ranging accuracy of ±1cm. With f / 1.4 large aperture technology, it is suitable for daytime scenarios and can collect real-time 2D image information of the road where the autonomous vehicle is located, including dynamic vehicle information on the road and surrounding static traffic facility information.

[0032] The roadside lidar is also deployed above the roadside poles. It adopts solid-state lidar with a ranging accuracy of ±1cm, an ultra-high angular resolution of 0.05°, and can output high-density pure point cloud information of 2 million points / second. It supports multi-echo detection and visual fusion technology, is suitable for all-weather scenarios, and can collect 3D point cloud information in the road environment in real time.

[0033] The edge computing unit receives road environment information collected by roadside cameras and roadside LiDAR, processes the 2D images and 3D point clouds using a visual perception fusion algorithm, and constructs an ultra-high-definition visual perception model as the ground truth model of the perception system. This ground truth model is then sent to the cloud platform. This model accurately reflects the actual state and positional relationships of dynamic vehicles and static traffic facilities in the current road environment, serving as the basis for the cloud platform to judge the accuracy of perception fusion for autonomous vehicles.

[0034] Meteorological information is collected by roadside meteorological sensors, including at least one of rain, snow, fog, dust storms, and clear weather. The roadside meteorological sensors are positioned above roadside poles to collect real-time meteorological data about the road environment in which the autonomous vehicle is located, and transmit this information to a cloud platform. This provides an environmental basis for subsequent sensor performance evaluation and arbitration ratio selection.

[0035] S2. Acquire the perception information and data quality parameters of each sensor in the vehicle perception system.

[0036] The sensors in the vehicle-mounted perception system include vehicle-mounted cameras, vehicle-mounted millimeter-wave radar, and vehicle-mounted lidar. The vehicle-mounted cameras are positioned in front of the autonomous vehicle to acquire real-time image perception information of the area in front of the vehicle; the vehicle-mounted millimeter-wave radar is positioned in front of the autonomous vehicle to acquire real-time millimeter-wave perception information of the area in front of the vehicle; and the vehicle-mounted lidar is positioned in front of the autonomous vehicle to acquire real-time point cloud perception information of the area in front of the vehicle. Data quality parameters, including image sharpness coefficients, are uploaded to the cloud platform along with the perception information from each sensor via the vehicle-mounted communication unit and the roadside communication unit. The image sharpness coefficients for each sensor are shown in Table 1.

[0037] Table 1 Image sharpness coefficients for each sensor

[0038] S3. Retrieve the influence coefficients of each sensor corresponding to the meteorological information, and calculate the perception quality score of each sensor in combination with the data quality parameters.

[0039] The calculation process for the perception quality score is as follows: the product of the influence coefficient of each sensor under the current meteorological information and the image sharpness coefficient of the corresponding sensor is used as the perception quality score of each sensor. The cloud platform pre-stores the influence coefficients of each sensor under different meteorological information, as shown in Table 2; the image quality scores of each sensor are shown in Table 3.

[0040] Table 2. Influence coefficients of different meteorological information for each sensor

[0041] Table 3 Image quality scores for each sensor

[0042] By pre-setting differentiated influence coefficients for each sensor under different weather conditions in the cloud platform and combining them with the sensor image clarity coefficients uploaded in real time by the vehicle controller, the perception quality score of each sensor is quantified using a product operation. This organically integrates the objective influence of the external weather environment with the real-time data quality of the sensor itself, thereby accurately identifying the sensor with the strongest actual perception capability under the current weather and sensor conditions as the optimal sensor. This provides a scientific and reliable decision-making basis for the subsequent directional selection of arbitration ratios. Millimeter-wave radar automatically obtains the highest score in rain, snow, and fog; lidar automatically obtains the highest score in sandstorm weather; and the camera automatically obtains the highest score in clear weather conditions. This fully conforms to the physical characteristics of different sensors under different weather conditions, significantly improving the objectivity and accuracy of optimal sensor selection.

[0043] S4. Determine the optimal sensor for the current road environment based on the perception quality score. The perception quality score includes an image quality score.

[0044] Specifically, this includes: comparing the perception quality scores of each sensor and determining the sensor with the highest score as the optimal sensor; when multiple sensors have the same highest score, determining the optimal sensor according to the preset priority of vehicle camera over vehicle LiDAR, which in turn takes precedence over vehicle millimeter-wave radar, so that the arbitration system can maintain a definite output under boundary or extreme conditions, avoiding situations where decisions cannot be made or decision results are frequently switched due to equal scores, thus ensuring the stability and consistency of system operation.

[0045] S5. Based on the optimal sensor, select candidate arbitration ratios from the preset perception fusion arbitration ratio case library.

[0046] Specifically, from the perception fusion arbitration ratio case library, a preset number of cases with the highest weight ratio of the corresponding optimal sensor are selected as candidate arbitration ratios. The preset number can be set as needed. In this embodiment, the cloud platform stores 10 perception fusion arbitration ratio cases, and four candidate arbitration ratios are selected. If the optimal sensor is millimeter-wave radar, the four cases with the highest weight ratio of millimeter-wave radar are selected; if the optimal sensor is lidar, the four cases with the highest weight ratio of lidar are selected; if the optimal sensor is a camera, the four cases with the highest weight ratio of camera are selected. The specific weight allocation of the 10 perception fusion arbitration ratio cases is shown in Table 4.

[0047] Table 4. Perceptual Fusion Arbitration Ratio Case Weighting Table

[0048] By pre-setting 10 cases of perception fusion arbitration ratios covering different weight combinations of each sensor, and selecting the case with the highest weight ratio of the optimal sensor as the candidate arbitration ratio, the scope of arbitration ratio optimization can be quickly reduced from all cases to a few candidate cases that are highly matched with the characteristics of the current optimal sensor. While ensuring that the candidate schemes fully cover the dominant position of the optimal sensor, the amount of computation for subsequent matching degree comparison is greatly reduced, and the real-time performance and efficiency of cloud arbitration are significantly improved.

[0049] S6. Simulate and fuse the perceived information using each candidate arbitration ratio, and compare the matching degree of each fusion result with the true value model of the perception system to determine the optimal arbitration ratio.

[0050] Specifically, the cloud platform uses the sensor weights defined by the selected candidate arbitration ratios to perform simulated fusion calculations on the actual perception information of the vehicle-mounted camera, vehicle-mounted millimeter-wave radar, and vehicle-mounted lidar uploaded by the vehicle controller, generating corresponding fusion results. Then, the fusion results corresponding to each candidate arbitration ratio are matched with the true value model of the perception system constructed by the edge computing unit. For example, the overlap or deviation of the features of the two are calculated through image matching algorithms, and the candidate arbitration ratio with the highest matching degree is determined as the optimal arbitration ratio. As shown in Table 5, under rain, snow, and fog conditions, millimeter-wave radar is the optimal sensor. Cases 8, 6, 7, and 1 (high proportion of millimeter-wave radar) were selected as candidate arbitration ratios for image matching comparison, and Case 1, with the highest matching degree, was determined as the optimal arbitration ratio. Under sandstorm conditions, lidar is the optimal sensor. Cases 9, 5, 4, and 1 (high proportion of lidar) were selected as candidate arbitration ratios for image matching comparison, and Case 5, with the highest matching degree, was determined as the optimal arbitration ratio. Under clear weather conditions, cameras are the optimal sensor. Cases 10, 2, 3, and 1 (high proportion of camera) were selected as candidate arbitration ratios for image matching comparison, and Case 3, with the highest matching degree, was determined as the optimal arbitration ratio. This optimal arbitration ratio is the fusion weight allocation scheme that best produces perception results consistent with the real scene under the current weather conditions and the current vehicle sensor status.

[0051] Table 5 Comparison of Optimal Arbitration Ratios

[0052] By simulating and fusing the perception information of each candidate arbitration ratio, and comparing each fusion result with the true value model of the perception system constructed by the roadside equipment, the arbitration ratio with the highest matching degree is selected as the optimal arbitration ratio. This ensures that the final arbitration ratio scheme does not depend solely on the theoretical weight, but has been verified by actual measurement comparison with the real environment benchmark. This ensures the optimality of the selected arbitration ratio in actual fusion effect, effectively guaranteeing the high degree of consistency between the fusion result of the vehicle perception system and the real road environment, and fundamentally improving the accuracy of perception fusion and the safety of autonomous vehicle decision-making.

[0053] S7. The optimal arbitration ratio is sent to the vehicle controller for fusion calculation of its on-board perception system.

[0054] Specifically, the cloud platform sends the identifier of the optimal arbitration ratio to the vehicle controller through the roadside communication unit and the vehicle communication unit. After receiving the optimal arbitration ratio, the vehicle controller performs weighted fusion calculation on the actual perception data of the vehicle camera, vehicle millimeter-wave radar and vehicle lidar according to the weight defined by the arbitration ratio, and constructs the final perception fusion model. Then, it controls the autonomous vehicle to make better driving decisions and control based on the fusion model. It can adaptively adjust the perception fusion model to be more suitable for different weather conditions for perception prediction and decision control, so that the autonomous driving system can more accurately judge the road environment and improve the safety of autonomous vehicles.

[0055] By distributing arbitration results in the form of the optimal arbitration ratio instead of transmitting complete weight data or fusion model parameters, the amount of data transmitted can be significantly reduced, communication bandwidth usage and transmission latency can be decreased, and the real-time performance of vehicle-side reception and execution can be improved. This enables autonomous vehicles to respond more quickly to changes in weather conditions and complete adaptive switching of perception fusion models.

[0056] The present invention also provides a perception fusion arbitration system based on vehicle-to-everything (V2X) communication for executing the above-described method. This perception fusion arbitration system includes a truth model acquisition module, a vehicle controller, a cloud platform, and a distribution module.

[0057] The vehicle controller is located at the bottom of the autonomous vehicle and is used to acquire the perception information and data quality parameters of each sensor in the onboard perception system.

[0058] The cloud platform retrieves the influence coefficients of each sensor corresponding to meteorological information and calculates the perception quality score of each sensor based on data quality parameters. The optimal sensor for the current road environment is determined based on the perception quality score. Based on the optimal sensor, candidate arbitration ratios are selected from a pre-defined perception fusion arbitration ratio case library. Each candidate arbitration ratio is used to perform simulated fusion calculations on the perceived information, and the fusion results are compared with the true model of the perception system to determine the optimal arbitration ratio. The cloud platform internally stores tables of sensor influence coefficients corresponding to different meteorological information and a perception fusion arbitration ratio case library.

[0059] The distribution module sends the optimal arbitration ratio to the vehicle controller for fusion calculation of its onboard perception system. The perception fusion arbitration system also includes a roadside communication unit (RSU) and an onboard communication unit. The RSU is located above road poles and transmits and receives data via 4G / 5G wireless communication. The onboard communication unit is located at the bottom of the autonomous vehicle and uses an onboard unit (OBU). The vehicle controller sends perception information and data quality parameters to the onboard communication unit, which in turn sends the perception information and data quality parameters to the cloud platform via the roadside communication unit. The roadside and onboard communication units interact wirelessly, forming a bidirectional data transmission link between the cloud platform and the vehicle controller. The roadside RSU receives instructions from the cloud platform and sends them wirelessly to the onboard OBU, while simultaneously receiving perception information and data quality parameters from the vehicle controller uploaded by the onboard OBU and sending them to the cloud platform.

[0060] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A perception fusion arbitration method based on vehicle-to-everything (V2X) communication, characterized in that, include: Acquire the true value model of the road environment perception system and meteorological information; Acquire the sensing information and data quality parameters of each sensor in the vehicle sensing system; The influence coefficients of each sensor corresponding to the meteorological information are retrieved, and the perception quality score of each sensor is calculated in combination with the data quality parameters. Based on the perception quality score, the optimal sensor for the current road environment is determined. Based on the optimal sensor, candidate arbitration ratios are selected from a preset case library of perception fusion arbitration ratios; each candidate arbitration ratio is used to simulate and fuse the perception information, and the matching degree of each fusion result is compared with the true value model of the perception system to determine the optimal arbitration ratio. The optimal arbitration ratio is sent to the vehicle controller for fusion calculation of its onboard perception system.

2. The perception fusion arbitration method based on vehicle-to-everything (V2X) as described in claim 1, characterized in that: The ground truth model of the perception system for acquiring the road environment specifically includes: Acquire 2D perception information from roadside cameras and 3D point cloud information from roadside lidar; Visual perception fusion is performed on the 2D perception information and 3D point cloud information to construct a truth model of the perception system.

3. The perception fusion arbitration method based on vehicle-to-everything (V2X) as described in claim 1, characterized in that: The meteorological information is collected by roadside meteorological sensors and includes at least one of rain, snow, fog, dust, and clear weather.

4. The perception fusion arbitration method based on vehicle-to-everything (V2X) as described in claim 1, characterized in that: The data quality parameters include the image sharpness coefficient; the perceived quality score includes the image quality score. The calculation process for the image quality score is as follows: The image quality score for each sensor is calculated by multiplying the influence coefficient of each sensor under the current meteorological information with the image sharpness coefficient of the corresponding sensor.

5. The perception fusion arbitration method based on vehicle-to-everything (V2X) as described in claim 1, characterized in that: The sensors in the vehicle-mounted perception system include vehicle-mounted cameras, vehicle-mounted millimeter-wave radar, and vehicle-mounted lidar.

6. The perception fusion arbitration method based on vehicle-to-everything (V2X) as described in claim 5, characterized in that: The optimal sensor for the current road environment is determined based on the perceived quality score, specifically including: The sensor with the highest perception quality score is determined as the optimal sensor by comparing the perception quality scores of each sensor. When multiple identical highest scores exist, the optimal sensor is determined according to the priority order of vehicle-mounted camera, vehicle-mounted LiDAR, and vehicle-mounted millimeter-wave radar.

7. The perception fusion arbitration method based on vehicle-to-everything (V2X) as described in claim 5, characterized in that: Candidate arbitration ratios are selected from a pre-defined perceptual fusion arbitration ratio case library, specifically including: The perception fusion arbitration comparison case library contains multiple arbitration comparison cases, and each arbitration comparison case defines the weight ratio of the vehicle camera, vehicle lidar and vehicle millimeter-wave radar in the perception system fusion calculation; From the perception fusion arbitration ratio case library, a preset number of cases with the highest corresponding optimal sensor weight ratio are selected as the candidate arbitration ratio.

8. A perception fusion arbitration system based on vehicle-to-everything (V2X) communication, characterized in that, The method for executing the vehicle-to-everything (V2X) based perception fusion arbitration method according to any one of claims 1-7 includes: The truth model acquisition module is used to acquire the truth model of the road environment perception system and meteorological information; The vehicle controller is used to acquire the sensing information and data quality parameters of each sensor in the vehicle sensing system. The cloud platform is used to retrieve the influence coefficients of each sensor corresponding to the meteorological information, and calculate the perception quality score of each sensor in combination with the data quality parameters, and determine the optimal sensor under the current road environment based on the perception quality score. Based on the optimal sensor, candidate arbitration ratios are selected from a preset case library of perception fusion arbitration ratios; each candidate arbitration ratio is used to simulate and fuse the perception information, and the matching degree of each fusion result is compared with the true value model of the perception system to determine the optimal arbitration ratio. The distribution module distributes the optimal arbitration ratio to the vehicle controller for fusion calculation of its on-board perception system.

9. The vehicle-to-everything (V2X) based perception fusion arbitration system according to claim 8, characterized in that: The truth model acquisition module includes: Roadside cameras are installed above roadside poles to collect road environment information; Roadside lidar is deployed above roadside poles to collect road environment information; Edge computing units are used to receive road environment information collected by roadside cameras and roadside lidar, and to build a true value model of the perception system; Roadside weather sensors are used to collect weather information.

10. The vehicle-to-everything (V2X) based perception fusion arbitration system according to claim 8, characterized in that: The perception fusion arbitration system also includes a roadside communication unit and an in-vehicle communication unit. The vehicle controller sends the perception information and data quality parameters to the in-vehicle communication unit. The vehicle-mounted communication unit sends the perceived information and data quality parameters to the cloud platform through the roadside communication unit.

Images