Reighting fusion detection method and system based on multi-agent system and edge calculation

By adopting the Chord loop routing table data transmission protocol in multi-agent system and edge computing, the problems of slow data transmission and low processing efficiency in radar and video fusion perception technology are solved, and real-time sharing of traffic events and vehicle information is realized and fast fusion calculations improve the timely distribution efficiency of analysis results.

CN119091609APending Publication Date: 2024-12-06INSPUR COMM TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411038271.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing radar and video fusion perception technologies have problems in the field of road safety, such as slow data transmission, low data processing efficiency and inability to be synchronized in time, which affects the real-time processing and distribution of analysis results.

Method used

Using the Lei Vision Fusion Detection Method based on multi-agent system and edge computing, data is collected, preprocessed and data transmission is carried out in real time through edge computing nodes, and low-latency data communication is achieved using Chord loop routing table, and in multi-agent system fusion calculation and result preservation are performed in.

Benefits of technology

Real-time sharing and rapid fusion calculation of traffic events and vehicle information is realized, data transmission delay is reduced, and analysis results are distributed to the backend center in a timely manner, improving processing efficiency and accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091609A_ABST
    Figure CN119091609A_ABST
Patent Text Reader

Abstract

The invention discloses a multi-agent system and edge computing-based thunder-vision fusion detection method and system, belongs to the technical field of edge computing, and aims to solve the technical problem of how to accurately and efficiently process and analyze thunder-vision fusion data and timely distribute processing results. Comprising the following steps that each edge computing node in the multi-agent system is regarded as an Agent node, the Agent nodes form a Chord ring clockwise from small to large based on identifiers of the Agent nodes, and each Agent node maintains a routing table; and addressing the target Agent node based on the routing table, searching the Agent node which is stored in the routing table of the target Agent node and is closest to the Agent node, downloading related traffic events and vehicle passing information from the searched Agent node, and carrying out fusion calculation to obtain an analysis result.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and specifically to a radar-vision fusion detection method and system based on a multi-agent system and edge computing. Background Art

[0002] With the development of technology, radar and video fusion perception technology is increasingly used in the field of road safety. This technology collects road traffic information in real time by collecting video and radar data, and then processes and analyzes the data on the backend server to grasp the road traffic conditions and vehicle-related information.

[0003] However, the existing analysis mode often has problems such as slow data transmission, low data processing efficiency, and inability to synchronize data information in a timely manner. These problems affect the real-time processing and analysis of radar-visual fusion data, and the real-time distribution of the processing and analysis results to other traffic areas.

[0004] How to accurately and efficiently process and analyze radar-vision fusion data and distribute the processing results in a timely manner is a technical problem that needs to be solved. Summary of the invention

[0005] The technical task of the present invention is to address the above shortcomings and provide a radar-vision fusion detection method and system based on a multi-agent system and edge computing to solve the technical problem of how to accurately and efficiently process and analyze radar-vision fusion data and distribute the processing results in a timely manner.

[0006] In a first aspect, the present invention provides a radar-visual fusion detection method based on a multi-agent system and edge computing, characterized in that it is applied between edge computing nodes of the multi-agent system and comprises the following steps:

[0007] Data collection: A smart traffic perception system is built based on radar and video fusion technology. Each edge computing node collects traffic events and vehicle traffic information in real time based on radar and video resources.

[0008] Data preprocessing: Each edge settlement node performs data preprocessing on the collected traffic events and vehicle communication information, stores the collected full data of traffic events and vehicle traffic as storage data locally, extracts a predetermined number of traffic events and vehicle communication parameters from the traffic events and vehicle traffic information as transmission data and sends them to other edge computing nodes adjacent to it;

[0009] Data communication transmission: Each edge computing node in the multi-agent system is regarded as an agent node. The agent nodes form a chord ring clockwise from small to large based on their identifiers. Each agent node maintains a routing table, which records the predecessor and successor nodes of the agent node. The agent nodes are addressed based on the routing table.

[0010] Data analysis: For the Agent node that needs to perform fusion calculation, the Agent node obtains the traffic events and vehicle traffic information stored locally and sent by the neighboring Agent nodes to perform fusion calculation to obtain the analysis result, or, based on the routing table, addresses the target Agent node, finds the Agent node closest to the Agent node stored in the routing table of the target Agent node, downloads the relevant traffic events and vehicle traffic information from the found Agent node, performs fusion calculation to obtain the analysis result;

[0011] Data preservation: The analysis results are saved to the event database and pushed to the backend center for storage.

[0012] Preferably, the vehicle traffic information collected based on radar includes distance parameters, speed parameters, and angle parameters of the vehicle, and the vehicle traffic information collected through video resources includes vehicle appearance information, vehicle color information, and license plate information;

[0013] Correspondingly, the stored data includes the edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, vehicle capture time, vehicle capture picture, vehicle capture video, radar capture vehicle distance parameter, radar capture vehicle speed parameter and radar capture angle parameter;

[0014] The transmitted data includes the edge computing node location number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, and vehicle capture time.

[0015] As a preference, for each Agent node's routing table, the length of the routing table is m, and for an Agent node with an identifier of n, the i-th item in the routing table stores the number of nodes in (n+2 i-1 )mod2 m Then the first Agent node s on the ring, s = sucdessor (n + 2 i-1 ), Agent node s is called the i-th router of Agent node n;

[0016] Each Agent node periodically checks the health status of its predecessor and successor nodes.

[0017] Preferably, the Agent node pushes the analysis results to the background center for storage when resources are idle.

[0018] In a second aspect, the present invention provides a radar-vision fusion detection system based on a multi-agent system and edge computing, which is used to implement radar-vision fusion detection by using a radar-vision fusion detection method based on a multi-agent system and edge computing as described in any one of the first aspects, including a data acquisition module, a data preprocessing module, a data communication transmission module, a data analysis module, and a data storage module;

[0019] The data acquisition module is used to build a smart traffic perception system based on radar and video fusion technology, and supports each edge computing node to perform the following: real-time collection of traffic events and vehicle traffic information based on radar and video resources;

[0020] The data preprocessing module is used to support each edge settlement node to perform the following: preprocess the collected traffic events and vehicle communication information, store the collected full data of traffic events and vehicle traffic as storage data locally, extract a predetermined number of traffic events and vehicle communication parameters from the traffic events and vehicle traffic information as transmission data and send them to other edge computing nodes adjacent to it;

[0021] The data communication transmission module is used to regard each edge computing node in the multi-agent system as an Agent node, and form a Chord ring clockwise from small to large based on its identifier. Each Agent node maintains a routing table, which records the predecessor and successor nodes of the Agent node. The Agent nodes are addressed based on the routing table;

[0022] For the Agent node that needs to perform fusion calculation, the data analysis module supports the Agent node to perform the following: obtain the traffic events and vehicle traffic information stored locally and sent by the neighboring Agent nodes to perform fusion calculation to obtain the analysis result, or address the target Agent node based on the routing table, find the Agent node closest to the Agent node stored in the routing table of the target Agent node, download the relevant traffic events and vehicle traffic information from the found Agent node to perform fusion calculation to obtain the analysis result;

[0023] The data preservation module supports the Agent node to perform the following: save the analysis results to the event database and push them to the backend center for storage.

[0024] Preferably, the vehicle traffic information collected based on radar includes distance parameters, speed parameters, and angle parameters of the vehicle, and the vehicle traffic information collected through video resources includes vehicle appearance information, vehicle color information, and license plate information;

[0025] Correspondingly, the stored data includes the edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, vehicle capture time, vehicle capture picture, vehicle capture video, radar capture vehicle distance parameter, radar capture vehicle speed parameter and radar capture angle parameter;

[0026] The transmitted data includes the edge computing node location number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, and vehicle capture time.

[0027] As a preference, for each Agent node's routing table, the length of the routing table is m, and for an Agent node with an identifier of n, the i-th item in the routing table stores the number of nodes in (n+2 i-1 )mod2 m Then the first Agent node s on the ring, s = sucdessor (n + 2 i-1 ), Agent node s is called the i-th router of Agent node n;

[0028] Each Agent node periodically checks the health status of its predecessor and successor nodes.

[0029] Preferably, the data storage module is used to support the Agent node to push the analysis results to the background center for storage when resources are idle.

[0030] The radar-vision fusion detection method and system based on multi-agent system and edge computing of the present invention have the following advantages:

[0031] 1. The data processing of collected traffic events and vehicle information is realized through the use of radar-vision fusion technology. At the same time, the collected information is pre-processed to extract key information. Agent nodes in the multi-agent network are used to be distributed and deployed on all edge computing nodes according to relevant transmission protocols, which can ensure real-time information sharing and reduce transmission delays.

[0032] 2. When data fusion calculation is required, the edge computing node can quickly obtain the required information data according to the communication transmission protocol, perform fusion calculation, and save the desired results to the background center. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] The present invention is further described below in conjunction with the accompanying drawings.

[0035] Figure 1 This is a flowchart of a radar-visual fusion detection method based on a multi-agent system and edge computing in Example 1;

[0036] Figure 2 This is a schematic diagram of data transmission between adjacent Agent nodes in a radar-vision fusion detection method based on a multi-agent system and edge computing in Example 1;

[0037] Figure 3 This is a schematic diagram of data transmission between Agent nodes based on a routing table in a radar-vision fusion detection method based on a multi-agent system and edge computing in Example 1. DETAILED DESCRIPTION

[0038] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments may be combined with each other.

[0039] The embodiments of the present invention provide a radar-vision fusion detection method and system based on a multi-agent system and edge computing, which are used to solve the technical problem of how to accurately and efficiently process and analyze radar-vision fusion data and distribute the processing results in a timely manner.

[0040] Embodiment 1:

[0041] The present invention provides a radar-vision fusion detection method based on a multi-agent system and edge computing, which is applied between edge computing nodes of the multi-agent system. The method includes five steps: data collection, data preprocessing, data communication transmission, data analysis and data storage.

[0042] Step S100 Data Collection: Build a smart traffic perception system based on radar and video fusion technology. Each edge computing node collects traffic events and vehicle traffic information in real time based on radar and video resources.

[0043] In this embodiment, the vehicle traffic information collected based on radar includes the vehicle's distance parameters, speed parameters, angle parameters, etc., and the vehicle traffic information collected through video resources includes vehicle appearance information, vehicle color information, and license plate information, etc.

[0044] Step S200 data preprocessing: Each edge settlement node performs data preprocessing on the collected traffic events and vehicle communication information, stores the collected full data of traffic events and vehicle traffic as storage data locally, extracts a predetermined number of traffic events and vehicle communication parameters from the traffic events and vehicle traffic information as transmission data and sends them to other edge computing nodes adjacent to it.

[0045] In this embodiment, the stored data includes the edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, vehicle capture time, vehicle capture picture, vehicle capture video, radar capture vehicle distance parameter, radar capture vehicle speed parameter and radar capture angle parameter. The full amount of data on traffic events and vehicle traffic is stored locally at the edge node, so that other edge computing terminals can request data reading and calling according to the information address location in the data communication transmission protocol when they have data call requirements.

[0046] In this embodiment, in order to minimize the delay of data transmission, only a small number of key parameters of traffic events and vehicle traffic are extracted to construct transmission data. The transmission data includes edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color and vehicle capture time. The Agent node and other Agent nodes of its neighbors transmit data according to the constructed communication transmission protocol.

[0047] Step S300 data communication transmission: Each edge computing node in the multi-agent system is regarded as an Agent node. The Agent nodes form a Chord ring clockwise from small to large based on their identifiers. Each Agent node maintains a routing table, which records the predecessor node and successor node of the Agent node. The Agent nodes are addressed based on the routing table.

[0048] In order to ensure the real-time sharing of traffic events and vehicle traffic related information, this embodiment constructs a low-latency data communication transmission protocol, and regards all edge computing nodes of the multi-agent system as Agent nodes of a Multi-Agent System. The two closest edge computing terminal nodes in the network, such as edge computing terminal nodes Agent1 and Agent2. Node Agent2 is the next node of Agent1. After Agent1 obtains information through the perception technology of radar and vision fusion, it processes the information according to step S200 and promptly transmits the processed brief information to its neighbor node Agent2.

[0049] At the same time, each node Agent maintains a routing table with a length of m. For the Agent node with identifier n, the i-th item in the routing table stores the number of nodes in (n+2i-1 )mod2 m Then the first Agent node s on the ring, s = sucdessor (n + 2 i-1 ), Agent node s is called the i-th route of Agent node n. The routing table records the predecessor and successor lists, which can quickly locate the predecessor and successor, and can periodically detect the health status of the predecessor and successor, so that each Agent node periodically detects the health status of its predecessor and successor nodes.

[0050] The stored successors are incremented in multiples of 2. The reason for taking the modulus is that the successor of the last node is the first few nodes. For example, the next node Agent of the largest node Agent is defined as the first node Agent.

[0051] The resource key is stored on the following Node: along the Chord ring, the first Node where hash(Node)>=hash(key) is called the successor of this key.

[0052] Given a key, follow the steps below to find which node Agent the corresponding resource is located on, that is, to find the successor of the key: (assuming the search is performed on node AgentN).

[0053] The processed traffic events and vehicle traffic summary information are transmitted to the corresponding nodes.

[0054] Step S400 data analysis: For the Agent node that needs to perform fusion calculation, the Agent node obtains the traffic events and vehicle traffic information stored locally and sent by the neighboring Agent nodes to perform fusion calculation to obtain the analysis result, or, based on the routing table, addresses the target Agent node, finds the Agent node closest to the Agent node stored in the routing table of the target Agent node, downloads the relevant traffic events and vehicle traffic information from the found Agent node to perform fusion calculation to obtain the analysis result.

[0055] In this embodiment, when an edge computing node collects traffic events and vehicle information, it needs to perform data fusion calculations with related nodes.

[0056] During execution, information transmitted from the previous node stored locally at the edge node is preferentially fused and calculated.

[0057] When the edge computing node Agent(x) wants to compare the traffic events and vehicle information of the edge computing node Agent(i), it performs addressing according to step S300, finds the node closest to the node Agent(x) stored in the Agent(i) routing table, downloads the relevant traffic events and vehicle information, and performs fusion calculation with the locally collected information.

[0058] Step S500: Data saving: The analysis results are saved to the event database and pushed to the backend center for storage.

[0059] In this embodiment, the corresponding processing results are saved in the event database and pushed to the background center for storage when the edge computing terminal resources are idle.

[0060] The method of this embodiment uses the radar-vision fusion technology to collect relevant traffic events and vehicle information at the roadside edge computing node, performs preprocessing at the edge computing node, extracts key information, transmits it between nodes according to the customized communication transmission protocol, and distributes and stores it. When new information is collected at the edge node and fusion calculation is to be performed, the relevant storage node is found according to the customized communication transmission protocol, and the relevant offline information is obtained for fusion calculation.

[0061] The method of this embodiment performs data preprocessing on the collected information and constructs a data structure suitable for transmission. In order to minimize the delay in data transmission, only a small number of key parameters of traffic events and vehicle traffic are extracted and transmitted according to the constructed communication transmission protocol.

[0062] In order to ensure the real-time sharing of traffic events and vehicle traffic related information, this embodiment constructs a low-latency data communication transmission protocol. When an edge computing node collects traffic events and vehicle information, it needs to perform data fusion calculation with related nodes.

[0063] Embodiment 2:

[0064] The present invention discloses a radar-vision fusion detection system based on a multi-agent system and edge computing, comprising a data acquisition module, a data preprocessing module, a data communication transmission module, a data analysis module and a data storage module.

[0065] The data acquisition module is used to build a smart traffic perception system based on radar and video fusion technology, and supports each edge computing node to perform the following: real-time collection of traffic events and vehicle traffic information based on radar and video resources.

[0066] In this embodiment, the vehicle traffic information collected based on radar includes the vehicle's distance parameters, speed parameters, angle parameters, etc., and the vehicle traffic information collected through video resources includes vehicle appearance information, vehicle color information, and license plate information, etc.

[0067] The data preprocessing module is used to support each edge settlement node to perform the following: perform data preprocessing on the collected traffic events and vehicle communication information, store the full amount of collected traffic events and vehicle traffic data locally as storage data, extract a predetermined number of traffic events and vehicle communication parameters from the traffic events and vehicle traffic information as transmission data and send them to other edge computing nodes adjacent to it.

[0068] In this embodiment, the stored data includes the edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, vehicle capture time, vehicle capture picture, vehicle capture video, radar capture vehicle distance parameter, radar capture vehicle speed parameter and radar capture angle parameter. The full amount of data on traffic events and vehicle traffic is stored locally at the edge node, so that other edge computing terminals can request data reading and calling according to the information address location in the data communication transmission protocol when they have data call requirements.

[0069] In this embodiment, in order to minimize the delay of data transmission, only a small number of key parameters of traffic events and vehicle traffic are extracted to construct transmission data. The transmission data includes edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color and vehicle capture time. The Agent node and other Agent nodes of its neighbors transmit data according to the constructed communication transmission protocol.

[0070] The data communication transmission module is used to regard each edge computing node in the multi-agent system as an Agent node, and to form a Chord ring clockwise from small to large based on the Agent nodes' identifiers. Each Agent node maintains a routing table, which records the predecessor and successor nodes of the Agent node. The Agent nodes are addressed based on the routing table.

[0071] In order to ensure the real-time sharing of traffic events and vehicle traffic related information, this embodiment constructs a low-latency data communication transmission protocol, and regards all edge computing nodes of the multi-agent system as Agent nodes of a Multi-Agent System. The two closest edge computing terminal nodes in the network, such as edge computing terminal nodes Agent1 and Agent2. Node Agent2 is the next node of Agent1. After Agent1 obtains information through the perception technology of radar and vision fusion, it processes the information according to step S200 and promptly transmits the processed brief information to its neighbor node Agent2.

[0072] Each node agent maintains a routing table with a length of m. For the agent node with identifier n, the i-th item in the routing table stores the number of nodes in (n+2i-1 )mod2 m Then the first Agent node s on the ring, s = sucdessor (n + 2 i -1 ), Agent node s is called the i-th route of Agent node n. The routing table records the predecessor and successor lists, which can quickly locate the predecessor and successor, and can periodically detect the health status of the predecessor and successor, so that each Agent node periodically detects the health status of its predecessor and successor nodes.

[0073] The stored successors in this embodiment are geometrically increased by multiples of 2. The reason for taking the modulus is that the successors of the last node are the first few nodes. For example, the next node Agent of the largest node Agent is defined as the first node Agent.

[0074] The resource key is stored on the following Node: along the Chord ring, the first Node where hash(Node)>=hash(key) is called the successor of this key.

[0075] Given a key, follow the steps below to find which node Agent the corresponding resource is located on, that is, to find the successor of the key: (assuming the search is performed on node AgentN).

[0076] The processed traffic events and vehicle traffic summary information are transmitted to the corresponding nodes.

[0077] For the Agent node that needs to perform fusion calculation, the data analysis module supports the Agent node to perform the following: obtain the traffic events and vehicle traffic information stored locally and sent by the neighboring Agent nodes for fusion calculation to obtain the analysis result, or address the target Agent node based on the routing table, find the Agent node closest to the Agent node stored in the routing table of the target Agent node, download the relevant traffic events and vehicle traffic information from the found Agent node for fusion calculation to obtain the analysis result.

[0078] In this embodiment, when an edge computing node collects traffic events and vehicle information, it needs to perform data fusion calculations with related nodes.

[0079] During execution, the data analysis module preferentially performs fusion calculation on the information transmitted from the previous node in the local storage of the edge node.

[0080] When the edge computing node Agent(x) wants to compare the traffic events and vehicle information of the edge computing node Agent(i), it performs addressing according to step S300, finds the node closest to the node Agent(x) stored in the Agent(i) routing table, downloads the relevant traffic events and vehicle information, and performs fusion calculation with the locally collected information.

[0081] The data preservation module supports the Agent node to perform the following: save the analysis results to the event database and push them to the backend center for storage.

[0082] In this embodiment, the data storage module saves the corresponding processing results to the event database, and pushes them to the background center for storage when the edge computing terminal resources are idle.

[0083] The system of this embodiment can execute the method disclosed in Example 1 to realize radar-visual fusion detection.

[0084] The present invention has been shown and described in detail above through the accompanying drawings and preferred embodiments. However, the present invention is not limited to these disclosed embodiments. Based on the above multiple embodiments, those skilled in the art can know that the means in the above different embodiments can be combined to obtain more embodiments of the present invention, and these embodiments are also within the protection scope of the present invention.

Claims

1. A radar-visual fusion detection method based on a multi-agent system and edge computing, characterized in that: Applied between edge computing nodes in a multi-agent system, including the following steps: Data collection: A smart traffic perception system is built based on radar and video fusion technology. Each edge computing node collects traffic events and vehicle traffic information in real time based on radar and video resources. Data preprocessing: Each edge settlement node performs data preprocessing on the collected traffic events and vehicle communication information, stores the collected full data of traffic events and vehicle traffic as storage data locally, extracts a predetermined number of traffic events and vehicle communication parameters from the traffic events and vehicle traffic information as transmission data and sends them to other edge computing nodes adjacent to it; Data communication transmission: Each edge computing node in the multi-agent system is regarded as an agent node. The agent nodes form a chord ring clockwise from small to large based on their identifiers. Each agent node maintains a routing table, which records the predecessor and successor nodes of the agent node. The agent nodes are addressed based on the routing table. Data analysis: For the Agent node that needs to perform fusion calculation, the Agent node obtains the traffic events and vehicle traffic information stored locally and sent by the neighboring Agent nodes to perform fusion calculation to obtain the analysis result, or, based on the routing table, addresses the target Agent node, finds the Agent node closest to the Agent node stored in the routing table of the target Agent node, downloads the relevant traffic events and vehicle traffic information from the found Agent node, performs fusion calculation to obtain the analysis result; Data preservation: The analysis results are saved to the event database and pushed to the backend center for storage.

2. The method for detecting by fusion of radar and vision based on multi-agent system and edge computing according to claim 1 is characterized in that: The vehicle traffic information collected based on radar includes the vehicle's distance parameters, speed parameters, and angle parameters. The vehicle traffic information collected through video resources includes vehicle shape information, vehicle color information, and license plate information. Correspondingly, the stored data includes the edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, vehicle capture time, vehicle capture picture, vehicle capture video, radar capture vehicle distance parameter, radar capture vehicle speed parameter and radar capture angle parameter; The transmitted data includes the edge computing node location number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, and vehicle capture time.

3. The method for radar-visual fusion detection based on multi-agent system and edge computing according to claim 1 is characterized in that: For each Agent node’s routing table, the length of the routing table is m. For an Agent node with identifier n, the i-th item in the routing table stores the number of nodes in (n+2 i-1 )mod2 m Then the first Agent node s on the ring, s = sucdessor (n + 2 i-1 ), Agent node s is called the i-th router of Agent node n; Each Agent node periodically checks the health status of its predecessor and successor nodes.

4. The method for radar-visual fusion detection based on multi-agent system and edge computing according to claim 3 is characterized in that: When resources are idle, the Agent node pushes the analysis results to the background center for storage.

5. A radar-visual fusion detection system based on a multi-agent system and edge computing, characterized in that: Used to realize radar-visual fusion detection by a radar-visual fusion detection method based on a multi-agent system and edge computing as described in any one of claims 1 to 4, comprising a data acquisition module, a data preprocessing module, a data communication transmission module, a data analysis module and a data storage module; The data acquisition module is used to build a smart traffic perception system based on radar and video fusion technology, and supports each edge computing node to perform the following: real-time collection of traffic events and vehicle traffic information based on radar and video resources; The data preprocessing module is used to support each edge settlement node to perform the following: preprocess the collected traffic events and vehicle communication information, store the collected full data of traffic events and vehicle traffic as storage data locally, extract a predetermined number of traffic events and vehicle communication parameters from the traffic events and vehicle traffic information as transmission data and send them to other edge computing nodes adjacent to it; The data communication transmission module is used to regard each edge computing node in the multi-agent system as an Agent node, and form a Chord ring clockwise from small to large based on its identifier. Each Agent node maintains a routing table, which records the predecessor and successor nodes of the Agent node. The Agent nodes are addressed based on the routing table; For the Agent node that needs to perform fusion calculation, the data analysis module supports the Agent node to perform the following: obtain the traffic events and vehicle traffic information stored locally and sent by the neighboring Agent nodes to perform fusion calculation to obtain the analysis result, or address the target Agent node based on the routing table, find the Agent node closest to the Agent node stored in the routing table of the target Agent node, download the relevant traffic events and vehicle traffic information from the found Agent node to perform fusion calculation to obtain the analysis result; The data preservation module supports the Agent node to perform the following: save the analysis results to the event database and push them to the backend center for storage.

6. The radar-visual fusion detection system based on multi-agent system and edge computing according to claim 5 is characterized in that: The vehicle traffic information collected based on radar includes the vehicle's distance parameters, speed parameters, and angle parameters. The vehicle traffic information collected through video resources includes vehicle shape information, vehicle color information, and license plate information. Correspondingly, the stored data includes the edge computing node point number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, vehicle capture time, vehicle capture picture, vehicle capture video, radar capture vehicle distance parameter, radar capture vehicle speed parameter and radar capture angle parameter; The transmitted data includes the edge computing node location number, event code, event time, vehicle unique ID, vehicle speed information, license plate information, license plate color, and vehicle capture time.

7. The radar-visual fusion detection system based on multi-agent system and edge computing according to claim 5 is characterized in that: For each Agent node’s routing table, the length of the routing table is m. For an Agent node with identifier n, the i-th item in the routing table stores the number of nodes in (n+2 i-1 )mod2 m Then the first Agent node s on the ring, s = sucdessor (n + 2 i-1 ), Agent node s is called the i-th router of Agent node n; Each Agent node periodically checks the health status of its predecessor and successor nodes.

8. The radar-visual fusion detection system based on multi-agent system and edge computing according to claim 5 is characterized in that: The data storage module is used to support the Agent node to push the analysis results to the background center for storage when resources are idle.