Traffic management method and system of auxiliary sensing robot, and electronic equipment

Abstract

The invention discloses a traffic management method and system for an auxiliary sensing robot and electronic equipment, and belongs to the technical field of traffic management and monitoring, and the method comprises the steps: carrying out the deployment decision of the auxiliary sensing robot according to a target region; performing robot deployment on the target area according to the robot deployment scheme; performing real-time traffic data acquisition on the target area according to the plurality of deployment robots to obtain a plurality of traffic monitoring data; traffic safety risk assessment is carried out according to the multiple traffic monitoring data; and sending the plurality of traffic safety risk coefficients to the plurality of deployment robots, and generating a traffic safety early warning signal. The technical problems of low traffic monitoring efficiency, untimely accident early warning and poor traffic safety management effect caused by unreasonable deployment, limited coverage and insufficient real-time traffic risk assessment of an auxiliary sensing robot in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the field of traffic management and monitoring technology, and in particular to a traffic management method, system and electronic equipment for an auxiliary perception robot. Background Art

[0002] With the acceleration of modern urbanization, the urban population is increasing worldwide, the number of vehicles has increased significantly, and urban traffic congestion and traffic safety issues are becoming increasingly prominent. Traditional traffic management methods mainly rely on infrastructure such as fixed cameras and roadside sensing equipment, which play a certain role in obtaining traffic data and identifying traffic risks.

[0003] However, in the existing traffic monitoring and management system, most of the work is still focused on the detection and alarm of road congestion, while the real-time capture and risk assessment of traffic accidents and dangerous situations are relatively insufficient. Therefore, how to use mobile devices to flexibly deploy in a wide area and combine intelligent data analysis methods to conduct real-time monitoring and early warning of traffic safety has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] This application aims to solve the problems of unreasonable deployment of auxiliary perception robots, limited coverage and insufficient real-time traffic risk assessment in the prior art by providing a traffic management method, system and electronic equipment for auxiliary perception robots, thereby leading to technical problems such as inefficient traffic monitoring, untimely accident warning and poor traffic safety management.

[0005] In a first aspect, the present application provides a traffic management method for an auxiliary perception robot, the method comprising making an auxiliary perception robot deployment decision based on a target area to obtain a robot deployment plan; deploying robots to the target area according to the robot deployment plan to determine a plurality of deployed robots; collecting real-time traffic data for the target area according to the plurality of deployed robots to obtain a plurality of traffic monitoring data; conducting a traffic safety risk assessment based on the plurality of traffic monitoring data to obtain a plurality of traffic safety risk coefficients; and sending the plurality of traffic safety risk coefficients to the plurality of deployed robots to generate a traffic safety warning signal.

[0006] In a second aspect, the present application provides a traffic management system for auxiliary perception robots, which includes a deployment decision module for making auxiliary perception robot deployment decisions based on a target area and obtaining a robot deployment plan; a robot deployment module for deploying robots to the target area according to the robot deployment plan and determining multiple deployed robots; a data acquisition module for collecting real-time traffic data for the target area based on the multiple deployed robots and obtaining multiple traffic monitoring data; a risk assessment module for performing traffic safety risk assessment based on the multiple traffic monitoring data and obtaining multiple traffic safety risk coefficients; and a warning signal generation module for sending the multiple traffic safety risk coefficients to the multiple deployed robots and generating a traffic safety warning signal.

[0007] In a third aspect, the present application provides an electronic device, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is used to execute the traffic management method for an auxiliary perception robot provided in the present application.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By adopting the traffic management method of auxiliary perception robots, including technical solutions for deployment decisions based on target areas, real-time traffic data collection, traffic safety risk assessment and warning signal generation, the problems of unreasonable deployment of auxiliary perception robots, limited monitoring coverage and insufficient real-time traffic risk assessment in the prior art, which led to technical problems such as low traffic monitoring efficiency, untimely accident warning and poor traffic safety management effect, the comprehensiveness and real-time nature of traffic monitoring, the accuracy and timeliness of accident warning, the overall efficiency of traffic safety management, the reduction of traffic accident rate and the optimization of the reasonable allocation and utilization of traffic resources are achieved.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A flow chart of a traffic management method for an auxiliary perception robot is provided for this application.

[0012] Figure 2 A structural schematic diagram of a traffic management method system for an auxiliary perception robot is provided for this application.

[0013] Figure 3A schematic diagram of the structure of an electronic device provided in this application.

[0014] Explanation of the accompanying drawings: deployment decision module 11, robot deployment module 12, data acquisition module 13, risk assessment module 14, warning signal generation module 15, processor 21, memory 22, input device 23, output device 24. DETAILED DESCRIPTION

[0015] The overall idea of ​​the technical solution provided by this application is as follows:

[0016] The embodiments of the present application provide a traffic management method, system and electronic device for an auxiliary perception robot. First, a deployment decision of the auxiliary perception robot is made based on the analysis of the target area, and a specific deployment plan is formulated. Then, multiple robots are deployed in the target area according to the plan to ensure wide coverage. The deployed robots collect traffic data in real time, and use this data to assess traffic safety risks and generate multiple risk coefficients. Finally, these risk coefficients are sent to each deployed robot to trigger a corresponding traffic safety warning signal.

[0017] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0018] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a traffic management method for an auxiliary perception robot, the method comprising:

[0019] Step S100: Make an auxiliary perception robot deployment decision according to the target area to obtain a robot deployment plan.

[0020] Specifically, the target area refers to a specific geographical area where traffic management or traffic monitoring is required in this technical solution, such as a city center, a specific highway section, an internal road network of a park, etc. An auxiliary perception robot refers to an intelligent terminal that has the ability to move autonomously or semi-autonomously and is equipped with perception equipment such as cameras, radars, sensors, etc., and can collect and analyze data about the environment in which it is located, such as a mobile patrol robot, an unmanned vehicle-mounted platform, and a drone.

[0021] First, it is necessary to conduct a comprehensive analysis of the geographic information, road structure and traffic status of the target area. To this end, geographic information system (GIS) software (such as ArcGIS or QGIS) can be used to import urban road network data and combine it with current or historical traffic flow data to identify key points such as traffic congestion sections and accident-prone areas.

[0022] Next, by building or calling an auxiliary perception robot deployment decision model, the deployment plan is deduced and optimized: after preliminarily obtaining the road topology information and traffic flow conditions in the target area, combined with the neural network model (such as using TensorFlow or PyTorch to learn a large number of historical road network deployment cases and traffic congestion distribution data), an "initial robot deployment plan" can be generated, such as deploying 2 robots at key intersections in the city center and 3 robots on the main roads.

[0023] Then, traffic flow prediction tools (such as LSTM, a time series prediction model based on deep learning, or rule-based traffic prediction algorithms) are used to predict the traffic pressure in the target area at different time periods, and these prediction results are used to modify the initial plan. For example, the main roads with the densest traffic during peak hours in the morning and evening may need to increase the frequency of robot patrols, while intersections with higher accident rates at night may need to set up special robot monitoring points.

[0024] On this basis, if certain roads are found to be accident-prone sections (this can be determined by reading past traffic accident data and performing cluster analysis or frequency evaluation), it is necessary to further increase the number of robots around these high-risk sections, or adjust their patrol trajectories, and integrate this information into the deployment plan.

[0025] Finally, a multi-objective optimization algorithm (such as genetic algorithm GA or particle swarm algorithm PSO) is used to comprehensively optimize the overall deployment plan of the robot (including quantity, location, patrol path, etc.), so as to take into account multiple aspects such as monitoring coverage, response time, and deployment cost. After the above, the "auxiliary perception robot deployment plan" is formed and output for subsequent implementation and verification.

[0026] This deployment decision-making method that integrates GIS data analysis, traffic flow prediction, machine learning models, and multi-objective optimization can greatly improve the deployment efficiency and monitoring coverage of auxiliary perception robots. While ensuring traffic safety in the target area, it can also effectively reduce monitoring costs and operation and maintenance burdens. The monitoring intensity at key intersections or accident-prone sections is enhanced, which helps to identify and warn of potential risks at an early stage, and improves the response speed and accuracy of traffic management departments in dealing with emergencies or peak congestion.

[0027] Step S200: Deploy robots to the target area according to the robot deployment plan, and determine a plurality of deployed robots.

[0028] Specifically, first of all, different types of auxiliary perception robots need to be prepared, including ground patrol robots, drones or unmanned vehicles equipped with cameras and sensors, etc., to meet the monitoring needs of different sections and different scenes. Then, combined with the locations or sections specified by the deployment plan, the specific location and responsible area of ​​each robot are marked through a visual map platform. At this point, scheduling management software (such as a multi-robot scheduling system developed based on Python or Java) can be used to record information such as robot number, operating period, patrol path or flight route, and send these parameters to the robot terminal.

[0029] During the deployment of robots, it is necessary to consider the network connection (such as 4G / 5G communication or Wi-Fi coverage) of each robot in the target area, the charging or energy supply location, and the safety of the parking or take-off and landing points. For example, for ground patrol robots, it is necessary to ensure that they can obtain charging piles or power access points near the deployment location; for drones, it is necessary to ensure that there is enough open space at the deployment point for take-off and landing, and comply with relevant airspace or flight control rules. After the deployment is completed, a remote monitoring system (such as a monitoring background based on a cloud platform) can be used to view the online status, positioning information, and perception data feedback of each robot to achieve dynamic tracking and management of the robot deployment effect.

[0030] By implementing the "Robot Deployment Plan" to actually deploy multiple perception robots, the original plan can be transformed from a theoretical level to practical implementation, thereby achieving more flexible, efficient and comprehensive traffic monitoring and safety management of the target area.

[0031] Step S300: collecting real-time traffic data of the target area according to the multiple deployed robots to obtain multiple traffic monitoring data.

[0032] Specifically, multiple traffic monitoring data refer to traffic data sets formed by aggregating or processing the perception results of all robots, which may include static data (such as road terrain) and dynamic data (such as vehicle speed, vehicle type, traffic density, etc.).

[0033] First, each robot will scan or monitor the surrounding traffic environment according to the type of sensors it carries and the set perception strategy. Next, each robot will transmit the collected data to the back-end traffic monitoring platform or cloud server in real time via a wireless network. During this process, the data will first undergo preliminary processing locally on the robot (such as video key frame extraction, simple target detection, etc.) to reduce network bandwidth pressure. Subsequently, the back-end system will integrate and correlate the raw data aggregated by all robots to form a multi-dimensional, multi-time "traffic monitoring data", including road speed distribution, vehicle queue length, intersection turning flow, accident detection and other information.

[0034] Through real-time traffic data collection, comprehensive coverage of different road sections, heights and angles can be achieved, improving the perception accuracy of mobility and sudden traffic conditions; multi-source and multi-type sensor data can be complementary and integrated for analysis, greatly improving detection accuracy (such as automatic identification of congestion, accidents, illegal parking, etc.).

[0035] Step S400: Performing a traffic safety risk assessment based on the plurality of traffic monitoring data to obtain a plurality of traffic safety risk coefficients.

[0036] Specifically, the traffic safety risk coefficient is the result of the traffic safety risk assessment, which is generally expressed in numerical values ​​or levels, reflecting the safety level or risk level of the current traffic environment. For example, the coefficient ranges from 0 to 1, with the higher the value, the greater the risk; or it can be divided into levels such as "low risk, medium risk, and high risk".

[0037] First, the cleaned and structured traffic monitoring data needs to be imported into the traffic safety risk assessment model. Using deep learning methods, the training data is input into the network for multiple rounds of iterative training, and the model performance (such as accuracy, recall rate, F1 value, etc.) is regularly evaluated on the test set. After the training is completed, a "traffic safety risk assessment model" will be obtained, which can output the corresponding risk score or risk coefficient based on the input real-time traffic monitoring data.

[0038] When new traffic monitoring data arrives, it is input into the trained model in the same feature format. The model evaluates these features based on the learned parameters and generates a set of "traffic safety risk factors", which are usually values ​​between 0 and 1, or directly outputs a certain risk level.

[0039] Through the process of "conducting traffic safety risk assessment based on the multiple traffic monitoring data", multi-source and dynamic traffic information can be converted into a quantified "traffic safety risk coefficient", providing a clear decision-making basis for traffic managers or intelligent transportation systems.

[0040] Step S500: Send the multiple traffic safety risk coefficients to the multiple deployed robots to generate a traffic safety warning signal.

[0041] Specifically, the traffic safety warning signal refers to the prompt or warning information sent by the system to the robot or other warning devices when the risk assessment results exceed the preset threshold (or meet certain trigger conditions) to remind or instruct them to take emergency or preventive measures. The signal can be in the form of sound and light alarms, voice broadcasts, remote message push, etc.

[0042] After the traffic safety risk assessment model outputs the risk coefficient, each coefficient is compared one by one. If the risk coefficient is ≥ a certain threshold value (such as 0.6, 0.7), or there is a significant upward trend, it is marked as "abnormal risk" or "high risk". More flexible methods such as rule judgment (if / else) or the introduction of fuzzy logic can be used to adapt to different road types or regional safety strategies. Secondary confirmation is performed on sections or time periods in a high-risk state, such as multiple exceeding of the threshold in a short period of time, or combined with the conditions of upstream and downstream roads, etc., to finally determine whether to trigger an alarm. If the alarm is confirmed to be triggered, the system will generate a corresponding "traffic safety warning signal" that can carry key information (high-risk section ID, current time period, risk level). At this time, a message queue (such as RabbitMQ, Kafka) or an IoT communication protocol (such as MQTT, HTTP REST) ​​can be used to send the warning signal to a specified robot group or all robots.

[0043] After receiving the warning signal, each robot performs the corresponding action according to its own configuration: If it is a ground patrol robot: activate the warning lights and speakers on the roof or body to warn vehicles or pedestrians; or immediately change the patrol route and quickly go to high-risk areas for emergency disposal and on-site verification. If it is a drone: turn on the warning broadcast, send out sound and light signals when patrolling in the air, and send the on-site high-definition video back to the dispatch center. If it is a vehicle-mounted platform: it can automatically generate a road diversion plan, or link with the roadside warning system (for example, it can control the roadside LED screen and electronic signs to issue warning reminders to passing vehicles). This process requires multi-machine collaboration to avoid repeated or chaotic scheduling. Emergency resources can be allocated and prioritized with the help of a multi-robot collaborative scheduling system (such as ROS-multi-machine version, custom cloud scheduling algorithm).

[0044] In the early warning state, the robot continuously monitors the corresponding road section and transmits data back, and the central platform continuously updates the risk factor. If the risk is reduced, the corresponding early warning signal can be closed or downgraded; if the risk continues to rise, more robots can be deployed or emergency services (rescue, traffic police, etc.) can be directly called.

[0045] By linking the traffic safety risk factor with the robot end, a real-time and efficient traffic management and safety early warning mechanism can be formed, which greatly reduces the pressure of manual monitoring and police input, and realizes the refinement and automation upgrade of urban traffic safety management.

[0046] Furthermore, an auxiliary perception robot deployment decision is made according to the target area to obtain a robot deployment plan, including: obtaining a regional road network of the target area; making an auxiliary perception robot deployment decision according to the regional road network to obtain a first robot deployment decision; making a traffic flow prediction based on the regional road network to obtain a regional traffic flow prediction result; adjusting the first robot deployment decision according to the regional traffic flow prediction result to obtain the robot deployment plan.

[0047] Specifically, the regional road network refers to the road topology and its ancillary information (including road location, number of lanes, traffic light distribution, intersection type, etc.) within the target area, which can usually be obtained and visualized with the help of geographic information system (GIS) data.

[0048] First, it is necessary to obtain the regional road network information in the target area, which can be achieved by calling GIS tools or map APIs to extract the road structure, node distribution and lane parameters in the target area. Then, the auxiliary perception robot deployment decision model is used to conduct a preliminary analysis of the road network data to generate the first deployment decision. This model can be trained through a neural network (such as PyTorch or TensorFlow framework) to determine the area or location where the robot is suitable for deployment based on features such as intersection connectivity, traffic density or terrain complexity, such as deploying three robots at key intersections and traffic nodes on the main road.

[0049] After obtaining the preliminary decision, the traffic flow prediction tools (such as the LSTM-based time series prediction model) are used to predict the traffic flow in the target area in different time periods, and the traffic flow change trend of each section during peak hours or specific time periods is obtained. For example, the model may predict that the traffic volume on a main road will surge from 7:30 to 9:30 in the morning, while the traffic volume on a remote section of the road will decrease at night but the probability of accidents will increase. Based on these prediction results, the preliminary decision is dynamically adjusted, such as increasing the number of robots in peak sections or extending their patrol time. In addition, genetic algorithms (GA) or particle swarm optimization algorithms (PSO) can be used to comprehensively optimize multiple constraints (such as monitoring coverage and robot energy consumption) to generate the final robot deployment plan.

[0050] The deployment method of assisted perception robots based on regional road network and traffic flow prediction can significantly improve the rationality and adaptability of robot deployment. On the one hand, the preliminary deployment decision provides a basic framework for rapid planning; on the other hand, the deployment is dynamically optimized through traffic flow prediction results, allowing robots to focus on monitoring in key periods and key areas.

[0051] Furthermore, an auxiliary perception robot deployment decision is made according to the regional road network to obtain a first robot deployment decision, including: training a robot deployment decision model according to a neural network; inputting the regional road network into the robot deployment decision model to generate the first robot deployment decision.

[0052] Specifically, it is first necessary to collect and organize the regional road network data of the target area, and extract elements such as road topology, intersection distribution, and road hierarchy. Next, establish or call a neural network model (such as a multi-layer perceptron or graph neural network based on TensorFlow or PyTorch), and use previous robot deployment plans and corresponding performance indicators (such as traffic accident reduction rate, congestion duration reduction rate, etc.) in similar road network environments as training samples for model training. By learning from a large number of training samples, the model can extract implicit rules such as road structure, key intersection connectivity, and vehicle flow characteristics, and output recommended deployment strategies for the area.

[0053] Specifically, the model input may include road traffic flow statistics, intersection topology, historical accident frequency, etc., and the output is the number of robots on the corresponding road section or node, patrol routes or monitoring priority distribution, etc. After training, the current regional road network data is input into this model to generate a "robot deployment first decision", such as deploying 1 mobile robot at each of the 5 high-accident rate intersections in the city center, and deploying 4 drones on 3 busy main roads for aerial monitoring. In this way, the distribution location and required number of robots can be preliminarily determined, so that they can be optimized in the subsequent steps in combination with traffic flow prediction or accident frequency evaluation.

[0054] By introducing neural networks to assist in the deployment of perception robots, we can fully explore the deep structure and traffic patterns of regional road networks with the support of massive historical data, and output more targeted and adaptive deployment plans. This helps to make preliminary decisions that take into account monitoring efficiency and deployment costs in the shortest time, and improve the intelligence and decision-making speed of urban traffic management.

[0055] Furthermore, the first robot deployment decision is adjusted according to the regional traffic flow prediction result to obtain the robot deployment plan, including: optimizing the first robot deployment decision according to the regional traffic flow prediction result to obtain the second robot deployment decision; collecting traffic accident record data of the regional road network to obtain a road network traffic accident record set; evaluating the traffic accident frequency of each road in the regional road network according to the road network traffic accident record set to obtain a traffic accident frequency evaluation result; adjusting the second robot deployment decision according to the traffic accident frequency evaluation result to obtain the robot deployment plan.

[0056] Specifically, the regional traffic flow prediction results refer to a series of values ​​or distributions obtained by analyzing, modeling and predicting the historical and real-time traffic flow data of the target area, usually including the trend of vehicle traffic changes and congestion duration at each road section or intersection in a specific time period. The first decision of robot deployment refers to the auxiliary perception robot deployment plan generated based on the existing road network analysis or preliminary modeling, which has not yet considered the correction link of dynamic traffic flow prediction factors or other safety hazards. The second decision of robot deployment refers to the new deployment plan obtained by optimizing or adjusting the "first decision of robot deployment" after taking the traffic flow prediction results into consideration, including adding robot monitoring or changing patrol routes during peak hours. The road network traffic accident record set refers to the data set related to traffic accidents that actually occurred in the target area, including the time, location, type of accident, loss, etc. The data can come from the accident records of the traffic police department, the insurance company's claims records or the public transportation database. The traffic accident frequency evaluation result refers to a set of indicators obtained by comprehensively evaluating the number and severity of accidents at each road section or intersection based on the road network traffic accident record set, which is used to measure the frequency of accidents on roads or sections.

[0057] First, the "first decision of robot deployment" is revised based on the regional traffic flow prediction results obtained previously to form a new second decision of robot deployment. Specifically: the traffic data of each road section or intersection in different time periods obtained based on the time series prediction algorithm (such as LSTM) or the macro traffic prediction model (such as the Four-Step model) is imported, and combined with the original deployment position of the robot, it is determined whether it is necessary to increase the deployment density or shorten the inspection cycle during high traffic periods.

[0058] Next, we collect traffic accident records from the road network, organize the time, location, and cause of traffic accidents in the past few years, and use statistical analysis software (such as pandas and NumPy under Python) or geographic information systems (ArcGIS and QGIS) to draw heat maps and perform cluster analysis, so as to calculate the accident frequency evaluation results of each road or intersection. If it is found that the accident density of a certain section is extremely high or accidents are frequent in a certain time period, the "second decision on robot deployment" needs to be adjusted again, and auxiliary perception robots should be added to the section, or more safety monitoring equipment should be deployed and the patrol frequency should be increased, so as to finally obtain a perfect "robot deployment plan".

[0059] By combining "traffic flow prediction" with "traffic accident frequency evaluation" and revising the deployment plan multiple times, accurate monitoring and deployment of roads in different time periods and risk levels can be achieved. On the one hand, the efficiency and coverage of auxiliary perception robots can be effectively improved; on the other hand, differentiated deployment based on accident-prone data can timely prevent and control potential risks and further reduce the probability of traffic accidents.

[0060] Furthermore, traffic safety risk assessment is performed based on the multiple traffic monitoring data to obtain multiple traffic safety risk coefficients, including: obtaining a traffic monitoring record set and a traffic safety risk assessment record set; performing data division based on the traffic monitoring record set and the traffic safety risk assessment record set to obtain a traffic safety risk assessment training set and a traffic safety risk assessment test set; performing training and testing based on the traffic safety risk assessment training set and the traffic safety risk assessment test set to generate a traffic safety risk assessment model; inputting the multiple traffic monitoring data into the traffic safety risk assessment model to output the multiple traffic safety risk coefficients.

[0061] Specifically, the traffic monitoring record set is a collection of traffic data collected by multiple deployed robots in the target area over a long period of time, including various original or pre-processed traffic-related information such as vehicle volume, average vehicle speed, traffic flow status, geographic location information, timestamp, etc. The traffic safety risk assessment record set is based on the record information of risk events such as historical accidents, traffic violations, congestion events, and the conditions under which these events occurred in the corresponding time and space, which can be regarded as historical data for labeling or rating "traffic safety risks".

[0062] First, it is necessary to obtain the traffic monitoring record set and the traffic safety risk assessment record set respectively. The traffic monitoring record set can be used by robots deployed on various roads to continuously collect traffic data for a period of time. The traffic safety risk assessment record set summarizes the traffic safety events and their risk characteristics that occurred in the corresponding time period or road section, including the type of accident, severity, time node, corresponding environmental factors, etc. Subsequently, it is necessary to synchronize or fuse these two types of data on the same time axis or spatial coordinates through data annotation or data matching to identify the association between a certain time and a certain road section and the corresponding risk event.

[0063] Next, according to the common machine learning or deep learning process, Python's pandas, NumPy and other data processing libraries can be used to split the two data sets, one as a training set for traffic safety risk assessment, and the other as a test set for traffic safety risk assessment. The training set is used to learn the model parameters, and the test set is used to verify the generalization ability of the model after the training is completed. In the formal training stage, deep learning methods such as multi-layer perceptron (MLP) or time series prediction network (LSTM) can be used, and neural networks can be built based on TensorFlow or PyTorch framework to achieve this. The model input usually includes feature vectors such as real-time traffic flow, road congestion index, weather conditions, and historical accident numbers, and the output is the risk score of the road section in the current or predicted period (such as a probability value of 0 to 1 or a multi-level classification). After the training is completed, the model will be evaluated on the test set to check its performance in indicators such as accuracy, recall, and F1 value to select the best model or the best hyperparameter configuration.

[0064] Finally, by inputting "multiple traffic monitoring data", i.e., new and real-time collected traffic condition characteristics, into the traffic safety risk assessment model obtained above, "multiple traffic safety risk coefficients" can be output. For example, at a certain intersection during the peak hour in the evening on a rainy day, the model calculates that its accident risk coefficient is 0.75, indicating a high risk, requiring enhanced early warning or dispatching more robot patrols. Through this process, the traffic management system can achieve efficient risk identification and real-time safety monitoring of complex road networks.

[0065] By combining traffic monitoring records with traffic safety risk assessment records and conducting scientific training and testing, and then using mature machine learning or deep learning algorithms to train a "traffic safety risk assessment model", the traffic safety risk coefficient can be output quickly and accurately in actual operation. This data-driven approach not only greatly improves the accuracy and timeliness of risk identification, but also enables continuous model iteration and optimization through new data, achieving adaptive learning and high-precision risk prediction for complex traffic environments.

[0066] Furthermore, the multiple traffic safety risk coefficients are sent to the multiple deployed robots to generate a traffic safety warning signal, including: determining whether the multiple traffic safety risk coefficients are greater than or equal to a traffic safety risk threshold; if any one of the multiple traffic safety risk coefficients is greater than or equal to the traffic safety risk threshold, obtaining an abnormal traffic safety risk coefficient; inputting the abnormal traffic safety risk coefficient into the traffic safety warning device of the multiple deployed robots to obtain the traffic safety warning signal.

[0067] Specifically, the traffic safety risk threshold refers to the numerical threshold set by the system to determine whether the traffic risk is within the "acceptable" range. When the risk factor equals or exceeds this threshold, it can be considered that the current state has a high traffic safety risk and requires warning or intervention.

[0068] The risk coefficient obtained by the traffic safety risk assessment model is transmitted to the management platform or cloud server for aggregation, and then judged and processed according to the following logic: Risk threshold judgment: The system will pre-set one or more traffic safety risk thresholds, such as 0.7 or 0.8, etc. The specific value can be determined based on historical data analysis or industry standards. The received risk coefficients are traversed or batch processed. If any coefficient is found to be ≥ the threshold, it can be determined that the data (or the road section, the time period) has an "abnormal traffic safety risk coefficient". Once it is detected that at least one risk coefficient exceeds or equals the threshold, the system will package the risk coefficient and related road section information, time information, etc. as "abnormal traffic safety risk data". In actual implementation, in order to prevent repeated alarms or misjudgments, a time limit mechanism can be set in the background. For example, if a road section has multiple small fluctuations in a short period of time but all exceed the threshold, it can be determined as the same continuous risk, and there is no need to repeatedly output multiple alarms.

[0069] When the system determines the "abnormal traffic safety risk coefficient", it immediately sends the coefficient and the corresponding location / time information to the "traffic safety warning devices of multiple deployed robots". Different robots can use different communication protocols, such as low-latency and highly reliable data transmission based on MQTT, HTTP REST or 5G IoT pipelines. After receiving the abnormal coefficient, the warning device will trigger the corresponding warning action according to the pre-configured response strategy: sound and light alarm, screen prompt, remote linkage: drones or vehicle-mounted platforms can start video feedback to assist the command center in obtaining on-site conditions and dispatching emergency personnel.

[0070] Through the mechanism of "judging risk factor → triggering warning", the system can make a coordinated response in the first time when high-risk or abnormal traffic conditions are discovered. On the one hand, it enables the robot end to quickly issue on-site sound and light warnings or information instructions to intervene in potential traffic accidents in time; on the other hand, it also provides visual and traceable decision-making support for the traffic management background, which facilitates the subsequent dispatch of emergency resources and post-event analysis.

[0071] Furthermore, real-time traffic data collection is performed on the target area according to the multiple deployed robots to obtain multiple traffic monitoring data, including: real-time traffic data collection is performed on the target area according to the multiple deployed robots to obtain multiple traffic data collection results; data cleaning is performed according to the multiple traffic data collection results to generate the multiple traffic monitoring data.

[0072] Specifically, traffic monitoring data refers to data sets that have been cleaned and structured, with higher accuracy and consistency, and can be directly used for subsequent traffic risk assessment, prediction and early warning analysis.

[0073] First, each robot uses sensors to sense the surrounding traffic environment and transmits back "multiple traffic data collection results". This result usually contains multiple types of raw information. Next, in order to ensure the high reliability of the data in the subsequent analysis phase, it is necessary to perform data cleaning operations on the "multiple traffic data collection results".

[0074] Specifically, check for data loss and format conflicts through scripts (such as Python scripts using pandas and NumPy libraries) or database ETL tools (such as Talend and Apache NiFi). If the robot is disconnected from the Internet in certain time periods, resulting in missing data, it is necessary to mark the missing values ​​or interpolate and fill in the gaps with data from similar time periods. Outlier processing: Eliminate or correct data with sudden jumps or sensor readings that deviate significantly from the normal range. Redundancy removal: If multiple robots repeatedly collect data on the same road section, they can be merged or filtered based on timestamps, geographic locations, or data similarities to reduce the storage and computational burden of duplicate records. Format unification and time alignment: Convert all data into a common format (such as a unified coordinate system and a unified timestamp granularity) so that the traffic flow, image information, etc. of each road section can be compared and correlated at the same time. Output structured results: Finally, "multiple traffic monitoring data" are formed for subsequent traffic safety risk assessment or real-time warning algorithm calls.

[0075] By systematically cleaning and merging the original traffic data, "multiple traffic data collection results" are converted into "multiple traffic monitoring data" with higher quality, more complete temporal and spatial alignment, and less noise.

[0076] In summary, the traffic management method of the auxiliary perception robot provided in the embodiment of the present application has the following technical effects:

[0077] 1. A full-process closed loop from the deployment of auxiliary perception robots to traffic safety warning has been built, which can realize real-time monitoring, risk assessment and warning linkage, effectively improve the traffic safety management efficiency in the target area, reduce the accident rate, and optimize resource allocation, providing a systematic solution for intelligent traffic management.

[0078] 2. Use neural network training to generate the first decision of robot deployment, so that the deployment plan can explore the deep connection between road network characteristics and traffic patterns. This technology generates optimization plans in an intelligent way, which helps to improve deployment efficiency and adaptability and reduce the blindness of traditional empirical deployment.

[0079] 3. Through the dual optimization strategy of traffic flow prediction and accident frequency evaluation, the pertinence and reliability of the robot deployment plan are improved. While ensuring the monitoring of high-risk sections, it also takes into account the overall optimization and achieves a balance between the reasonable allocation of resources and risk management.

[0080] Embodiment 2, based on the same inventive concept as the traffic management method of the auxiliary perception robot in the above embodiment, Figure 2 As shown, an embodiment of the present application provides a traffic management system for an auxiliary perception robot, the system comprising:

[0081] A deployment decision module 11 is used to make auxiliary perception robot deployment decisions according to the target area and obtain a robot deployment plan; a robot deployment module 12 is used to deploy robots to the target area according to the robot deployment plan and determine multiple deployed robots; a data acquisition module 13 is used to collect real-time traffic data for the target area according to the multiple deployed robots and obtain multiple traffic monitoring data; a risk assessment module 14 is used to perform traffic safety risk assessment based on the multiple traffic monitoring data and obtain multiple traffic safety risk coefficients; an early warning signal generation module 15 is used to send the multiple traffic safety risk coefficients to the multiple deployed robots and generate a traffic safety early warning signal.

[0082] Furthermore, the deployment decision module 11 is also used to perform the following steps: obtain the regional road network of the target area; make an auxiliary perception robot deployment decision based on the regional road network to obtain a first robot deployment decision; perform traffic flow prediction based on the regional road network to obtain a regional traffic flow prediction result; adjust the first robot deployment decision according to the regional traffic flow prediction result to obtain the robot deployment plan.

[0083] Furthermore, the deployment decision module 11 is also used to perform the following steps: training a robot deployment decision model based on a neural network; inputting the regional road network into the robot deployment decision model to generate the first robot deployment decision.

[0084] Furthermore, the deployment decision module 11 is also used to perform the following steps: optimize the first robot deployment decision according to the regional traffic flow prediction result to obtain the second robot deployment decision; collect traffic accident record data of the regional road network to obtain a road network traffic accident record set; evaluate the traffic accident frequency of each road in the regional road network according to the road network traffic accident record set to obtain a traffic accident frequency evaluation result; adjust the second robot deployment decision according to the traffic accident frequency evaluation result to obtain the robot deployment plan.

[0085] Furthermore, the risk assessment module 14 is also used to perform the following steps: obtain a traffic monitoring record set and a traffic safety risk assessment record set; perform data division according to the traffic monitoring record set and the traffic safety risk assessment record set to obtain a traffic safety risk assessment training set and a traffic safety risk assessment test set; perform training and testing according to the traffic safety risk assessment training set and the traffic safety risk assessment test set to generate a traffic safety risk assessment model; input the multiple traffic monitoring data into the traffic safety risk assessment model, and output the multiple traffic safety risk coefficients.

[0086] Furthermore, the warning signal generating module 15 is also used to perform the following steps: determine whether the multiple traffic safety risk coefficients are greater than or equal to the traffic safety risk threshold; if any one of the multiple traffic safety risk coefficients is greater than or equal to the traffic safety risk threshold, obtain an abnormal traffic safety risk coefficient; input the abnormal traffic safety risk coefficient into the traffic safety warning device of the multiple deployed robots to obtain the traffic safety warning signal.

[0087] Furthermore, the data acquisition module 13 is also used to perform the following steps: collecting real-time traffic data of the target area according to the multiple deployed robots to obtain multiple traffic data collection results; performing data cleaning according to the multiple traffic data collection results to generate the multiple traffic monitoring data.

[0088] Embodiment 3, Figure 3 The schematic diagram of the structure of the electronic device provided in the third embodiment of the present invention shows a block diagram of an exemplary electronic device suitable for implementing the implementation mode of the present invention. Figure 3 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention. Figure 3 As shown, the electronic device includes a processor 21, a memory 22, an input device 23 and an output device 24; the number of processors 21 in the electronic device can be one or more. Figure 3 Taking a processor 21 as an example, the processor 21, memory 22, input device 23 and output device 24 in the electronic device can be connected through a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0089] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0090] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A traffic management method using an auxiliary perception robot, characterized in that: The method comprises: Make auxiliary perception robot deployment decisions based on the target area and obtain the robot deployment plan; Deploy robots to the target area according to the robot deployment plan, and determine a plurality of deployed robots; Collecting real-time traffic data of the target area according to the multiple deployed robots to obtain multiple traffic monitoring data; Performing a traffic safety risk assessment based on the multiple traffic monitoring data to obtain multiple traffic safety risk coefficients; The multiple traffic safety risk coefficients are sent to the multiple deployed robots to generate traffic safety warning signals.

2. The method according to claim 1, characterized in that Make auxiliary perception robot deployment decisions based on the target area and obtain the robot deployment plan, including: Obtaining a regional road network of the target area; Making an auxiliary perception robot deployment decision according to the regional road network to obtain a first robot deployment decision; Performing traffic flow prediction based on the regional road network to obtain regional traffic flow prediction results; The first robot deployment decision is adjusted according to the regional traffic flow prediction result to obtain the robot deployment plan.

3. The method according to claim 2, characterized in that The auxiliary perception robot deployment decision is made according to the regional road network to obtain a first robot deployment decision, including: Based on the neural network, train the robot deployment decision model; The regional road network is input into the robot deployment decision model to generate the first decision of the robot deployment.

4. The method according to claim 2, characterized in that Adjusting the first robot deployment decision according to the regional traffic flow prediction result to obtain the robot deployment plan includes: Optimizing the first robot deployment decision according to the regional traffic flow prediction result to obtain a second robot deployment decision; Collecting traffic accident record data of the regional road network to obtain a road network traffic accident record set; Evaluate the traffic accident frequency of each road in the regional road network according to the road network traffic accident record set to obtain a traffic accident frequency evaluation result; The second robot deployment decision is adjusted according to the traffic accident frequency evaluation result to obtain the robot deployment plan.

5. The method according to claim 1, characterized in that Traffic safety risk assessment is performed based on the plurality of traffic monitoring data to obtain a plurality of traffic safety risk coefficients, including: Obtaining traffic monitoring record sets and traffic safety risk assessment record sets; Performing data division according to the traffic monitoring record set and the traffic safety risk assessment record set to obtain a traffic safety risk assessment training set and a traffic safety risk assessment test set; Performing training and testing based on the traffic safety risk assessment training set and the traffic safety risk assessment test set to generate a traffic safety risk assessment model; The plurality of traffic monitoring data are input into the traffic safety risk assessment model, and the plurality of traffic safety risk coefficients are output.

6. The method according to claim 1, characterized in that The method further comprises: sending the plurality of traffic safety risk coefficients to the plurality of deployed robots to generate a traffic safety warning signal, including: Determining whether the multiple traffic safety risk coefficients are greater than or equal to a traffic safety risk threshold; If any one of the multiple traffic safety risk coefficients is greater than or equal to the traffic safety risk threshold, obtaining an abnormal traffic safety risk coefficient; The abnormal traffic safety risk coefficient is input into the traffic safety warning device of the multiple deployed robots to obtain the traffic safety warning signal.

7. The method according to claim 1, characterized in that The multiple deployed robots collect real-time traffic data of the target area to obtain multiple traffic monitoring data, including: Collecting real-time traffic data of the target area according to the multiple deployed robots to obtain multiple traffic data collection results; Data cleaning is performed according to the plurality of traffic data collection results to generate the plurality of traffic monitoring data.

8. Traffic management system of auxiliary perception robot, characterized in that, A traffic management method for an auxiliary perception robot according to any one of claims 1 to 7, the system comprising: The deployment decision module is used to make auxiliary perception robot deployment decisions according to the target area and obtain the robot deployment plan; A robot deployment module, used to deploy robots to the target area according to the robot deployment plan, and determine a plurality of deployed robots; A data collection module, used for collecting real-time traffic data of the target area according to the multiple deployed robots to obtain multiple traffic monitoring data; A risk assessment module, used to perform traffic safety risk assessment based on the plurality of traffic monitoring data to obtain a plurality of traffic safety risk coefficients; The warning signal generating module is used to send the multiple traffic safety risk coefficients to the multiple deployed robots to generate a traffic safety warning signal.

9. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; Wherein, the processor is used to execute the traffic management method of the auxiliary perception robot as described in any one of claims 1 to 7.

Images