Intelligent emergency response method and system based on multi-modal traffic event detection

Abstract

This invention discloses an intelligent emergency response method and system based on multimodal traffic event detection, relating to the field of intelligent transportation technology. The method includes: acquiring technical parameters of traffic equipment and road environment parameters; configuring a multi-objective optimization model that meets the needs of urban emergency management, dynamically adjusting the activation sequence of emergency plans, the frequency of multimodal perception data collection, and edge computing resource allocation strategies; simulating the road network response efficiency and event recognition accuracy under different traffic event types, generating emergency response plans for multimodal traffic events; and deploying edge computing nodes and configuring traffic emergency control commands. This invention solves the technical problems of outdated emergency response mechanisms, rigid plans, low resource scheduling efficiency, and lack of dynamic simulation and verification of road network efficiency in existing technologies. It achieves the technical effects of improving the real-time performance, accuracy, and dynamic adaptability of traffic event emergency response, optimizing the utilization rate of edge computing resources, and enhancing the overall operational efficiency of the road network.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, specifically to an intelligent emergency response method and system based on multimodal traffic event detection. Background Technology

[0002] With the acceleration of urbanization and the surge in motor vehicle ownership, urban road traffic networks are facing unprecedented pressure. Traffic accidents, vehicle breakdowns, severe weather, road construction, and congestion caused by large events are random, sudden, and have a chain reaction effect. If they are not dealt with in a timely and effective manner, they can easily cause regional traffic congestion and even lead to secondary accidents, threatening traffic efficiency. Traditional traffic emergency response relies primarily on manual inspections, landline alarms, and limited closed-circuit television monitoring. When a traffic incident occurs, the command center typically receives the alarm, manually retrieves on-site video footage, assesses the incident's severity, and issues commands based on pre-set static emergency plans. However, this approach has several shortcomings in practical application. Incident perception suffers from blind spots and poor timeliness. Furthermore, existing emergency plans struggle to adapt to real-time traffic equipment conditions such as camera malfunctions and traffic signal failures, changes in road conditions like visibility and slipperiness, and the dynamic evolution of the incident, easily leading to a waste of emergency resources. In addition, during emergencies, massive amounts of multimodal sensing data from video, radar, and geomagnetic sensors flood the central platform, severely impacting the central server's computing power and network bandwidth. The current centralized processing architecture struggles to guarantee low-latency processing of critical data under extreme conditions, preventing the dynamic scheduling of edge computing resources to prioritize high-priority events and thus affecting the efficiency of control command issuance.

[0003] Therefore, current technologies suffer from technical problems such as outdated emergency response mechanisms, rigid contingency plans, low resource allocation efficiency, and a lack of dynamic simulation and verification of road network performance. Summary of the Invention

[0004] This application provides an intelligent emergency response method and system based on multimodal traffic incident detection, which solves the technical problems of outdated emergency response mechanisms, rigid plans, low resource scheduling efficiency, and lack of dynamic simulation and verification of road network performance in the existing technology. It achieves the technical effects of improving the real-time performance, accuracy, and dynamic adaptability of traffic incident emergency response, optimizing the utilization rate of edge computing resources, and enhancing the overall operational efficiency of the road network.

[0005] This application provides an intelligent emergency response method based on multimodal traffic event detection. The method includes: acquiring technical parameters of traffic equipment and road environment parameters; configuring a multi-objective optimization model that meets the needs of urban emergency management based on the technical parameters of traffic equipment and road environment parameters, dynamically adjusting the activation sequence of emergency plans, the frequency of multimodal perception data collection, and the edge computing resource allocation strategy; simulating the road network response efficiency and event recognition accuracy under different types of traffic events, and generating an emergency response plan for multimodal traffic events by combining the multi-objective optimization model; deploying edge computing nodes and configuring traffic emergency control commands using the emergency response plan.

[0006] In a possible implementation, a multi-objective optimization model that meets the needs of urban emergency management is configured based on the technical parameters of the traffic equipment and the road environment parameters. The method also includes setting a constraint range by taking the emergency response start delay, video analysis frame rate, radar point cloud update cycle, V2X message broadcast frequency, and edge AI inference load as decision variables. The constraint range includes the minimum emergency lane release time, the maximum information miss rate, and the upper limit of communication bandwidth.

[0007] Among possible implementations, the intelligent emergency response method based on multimodal traffic event detection includes: accessing a data interaction interface, uploading event confidence, vehicle trajectory anomaly, and affected area range, and constructing a digital twin of the traffic event; performing dynamic deduction based on the digital twin of the traffic event, and feeding back the decision variables for optimizing the multi-objective optimization model.

[0008] In a possible implementation, the emergency response scheme is used to configure traffic emergency control commands. The method includes: the traffic emergency control commands include emergency phase switching commands for traffic lights, emergency lane opening commands, and traffic diversion commands for surrounding intersections; based on the edge computing node, a synchronization mechanism is configured on the data interaction interface, the synchronization mechanism being used to correct the transmission delay of the traffic emergency control commands.

[0009] In one possible implementation, the intelligent emergency response method based on multimodal traffic incident detection includes: deploying link test anchors at key communication nodes, wherein the key communication nodes include data interaction interfaces and V2X message forwarding nodes; and triggering communication relay enhancement when the message packet loss rate within the key communication nodes is detected to be higher than a threshold value.

[0010] In one possible implementation, the intelligent emergency response method based on multimodal traffic incident detection includes: determining communication assurance priorities based on a traffic incident modal knowledge base and the dynamic inference results of the traffic incident digital twin; and dynamically adjusting the strength and timing of communication relay enhancements according to the communication assurance priorities.

[0011] Among possible implementations, the intelligent emergency response method based on multimodal traffic event detection includes: constructing a traffic event modal knowledge base, which stores emergency response time requirements and multimodal perception resource allocation strategies under the annotations of traffic accident mode, vehicle spontaneous combustion mode, and road collapse mode.

[0012] This application also provides an intelligent emergency response system based on multimodal traffic event detection. The system includes: a data acquisition module for acquiring technical parameters of traffic equipment and road environment parameters; a dynamic adjustment module for configuring a multi-objective optimization model that meets the needs of urban emergency management based on the technical parameters of traffic equipment and road environment parameters, and dynamically adjusting the activation sequence of emergency plans, the frequency of multimodal perception data collection, and the edge computing resource allocation strategy; a scheme generation module for simulating the road network response efficiency and event recognition accuracy under different traffic event types, and generating emergency response schemes under multimodal traffic events in combination with the multi-objective optimization model; and an instruction configuration module for deploying edge computing nodes and configuring traffic emergency control instructions using the emergency response scheme.

[0013] This application proposes an intelligent emergency response method and system based on multimodal traffic event detection. The method aims to acquire technical parameters of traffic equipment and road environment parameters; configure a multi-objective optimization model that meets the needs of urban emergency management; dynamically adjust the activation sequence of emergency plans, the frequency of multimodal sensing data collection, and the allocation strategy of edge computing resources; simulate the road network response efficiency and event recognition accuracy under different traffic event types; generate emergency response plans for multimodal traffic events; and deploy edge computing nodes and configure traffic emergency control commands. This addresses the technical problems in existing technologies, such as lagging emergency response mechanisms, rigid plans, low resource scheduling efficiency, and a lack of dynamic simulation and verification of road network efficiency. It achieves the technical effects of improving the real-time performance, accuracy, and dynamic adaptability of traffic event emergency response, optimizing the utilization rate of edge computing resources, and enhancing the overall operational efficiency of the road network. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0015] Figure 1 This is a schematic diagram of the intelligent emergency response method based on multimodal traffic incident detection provided in an embodiment of this application.

[0016] Figure 2 This is a schematic diagram of the structure of an intelligent emergency response system based on multimodal traffic incident detection provided in an embodiment of this application.

[0017] Figure labeling: Data acquisition module 10, dynamic adjustment module 20, scheme generation module 30, instruction configuration module 40. Detailed Implementation

[0018] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structure, features, and effects of the present invention.

[0019] This application provides an intelligent emergency response method based on multimodal traffic incident detection, such as... Figure 1 As shown, the method includes:

[0020] Step S100: Obtain the technical parameters of the traffic equipment and the road environment parameters.

[0021] Preferably, the technical parameters of traffic equipment refer to the working status, performance indicators, and operating data of traffic electromechanical equipment such as cameras, radars, and traffic signals deployed on the roadside and in the center. These include, but are not limited to, parameters of video surveillance equipment, such as online status, video bitrate, resolution, frame rate, lens shooting angle, pitch angle, and memory usage; parameters of radar detection equipment, such as radar transmission frequency, power, detection range, horizontal / vertical field of view, and target detection accuracy; parameters of traffic signal controllers, such as current timing scheme, signal working mode, countdown display status, and round-trip latency and packet loss rate for communication with the center; parameters of V2X roadside units, such as communication coverage, signal strength, signal-to-noise ratio, message broadcast frequency, and channel busy rate; and parameters of edge computing nodes, such as CPU / GPU / memory utilization, remaining storage capacity, and network uplink / downlink bandwidth usage. Road environmental parameters refer to data on the condition of road infrastructure and surrounding environmental conditions when a traffic incident occurs. These parameters include at least meteorological environmental parameters such as visibility, road surface temperature, wind speed and direction, and road surface condition; road infrastructure condition parameters such as lane closure status, road alignment, road surface smoothness, friction coefficient, on / off status of lighting facilities and illuminance; and traffic flow environmental parameters such as average vehicle speed, traffic flow, time occupancy, queue length, headway, and distance between vehicles.

[0022] Step S200: Based on the technical parameters of the traffic equipment and the road environment parameters, configure a multi-objective optimization model that meets the needs of urban emergency management, and dynamically adjust the activation sequence of the emergency plan, the frequency of multimodal perception data collection, and the edge computing resource allocation strategy.

[0023] Preferably, real-time traffic equipment technical parameters and road environment parameters are used as inputs and substituted into a multi-objective optimization model that meets the needs of urban emergency management for calculation. The output results are used to dynamically adjust three control variables: the activation sequence of emergency plans, the frequency of multimodal sensing data collection, and the edge computing resource allocation strategy. Specifically, urban emergency management needs refer to multiple quantitative urban management objectives that the model needs to meet simultaneously. For example, the timeliness objective is that the total delay from the occurrence of an event to the issuance of the first wave of control instructions should not exceed 20 seconds; the accuracy objective is that the accuracy rate of identifying event types (such as accidents and spontaneous combustion) should not be lower than 98%; the reliability objective is that the packet loss rate of emergency instructions during transmission should not exceed 2%; and the coverage objective is that the coverage rate of sensing data collection in the event-affected area should not be lower than 85%. Based on linear programming or reinforcement learning models, a multi-objective optimization model is configured according to multiple quantitative objectives to find the optimal solution under the premise of meeting all urban emergency management objectives.

[0024] Preferably, dynamically adjusting the activation sequence of emergency plans refers to automatically determining the time and order of activating different levels of emergency plans. This includes judging the severity of the event based on input data. For example, if the model input shows visibility below 50 meters and multiple vehicles experiencing a sudden drop in speed, it calculates "immediately activate the highest-level emergency plan, and activate traffic light red light control within 3 seconds, and activate traffic information screen guidance within 5 seconds." If it shows only one vehicle with hazard lights flashing and stopped, it calculates "activate the fourth-level emergency plan after 2 minutes, and initiate the video confirmation process first." Dynamically adjusting the multimodal perception data acquisition frequency refers to automatically adjusting the frequency of data transmission from different sensors such as cameras, radar, and weather stations. This includes dynamic scheduling based on the event's impact range and equipment load. For example, in the core area of ​​the event, it might instruct "camera frame rate to increase from 25fps to 50fps, and radar point cloud update cycle to shorten from 100ms to 50ms." In the peripheral area of ​​the event, to save bandwidth and computing resources, it might instruct "camera frame rate to decrease to 5fps, and only key frames to be transmitted." If the CPU load of edge nodes exceeds 90%, it might instruct a reduction in the data acquisition frequency of non-core areas. Dynamically adjusting edge computing resource allocation strategies refers to automatically allocating the computing power of roadside edge computing nodes such as smart cameras, RSUs, and edge servers to execute different computing tasks. This includes scheduling based on task priority and node computing power. For example, under normal circumstances, 60% of the computing power of an edge node is used for routine traffic flow statistics, and 30% is used for basic event detection. When an event occurs, it may be instructed to "redistribute 90% of the node's GPU resources to the event video analysis AI task and suspend non-critical log compression tasks." If edge computing node A has insufficient computing power, some video analysis tasks will be offloaded to the adjacent edge computing node B for processing.

[0025] Furthermore, step S200 also includes setting a constraint range by taking the emergency response activation delay, video analysis frame rate, radar point cloud update cycle, V2X message broadcast frequency, and edge AI inference load as decision variables; the constraint range includes the minimum emergency lane release time, the maximum information false alarm rate, and the upper limit of communication bandwidth.

[0026] Preferably, decision variables refer to parameters in the multi-objective optimization model that can be actively adjusted and controlled, including emergency response initiation delay, video analysis frame rate, radar point cloud update cycle, V2X message broadcast frequency, and edge AI inference load. Emergency response initiation delay measures the time elapsed between confirming a traffic incident and generating and issuing the first emergency control command, thus measuring the system's reaction speed. Video analysis frame rate refers to the number of image frames per second processed by AI analysis of the video stream; a higher frame rate allows for more continuous capture of event details but also consumes greater computational resources and bandwidth. Radar point cloud update cycle refers to the time between two complete scans by the radar equipment and the transmission of data to the system. The shorter the time interval between data transmissions, the more real-time the detection of vehicle position and speed, but the higher the amount of data generated and the higher the subsequent processing load. The V2X message broadcast frequency refers to the number of times the roadside unit sends V2X messages to surrounding vehicles per second. The higher the frequency, the faster the surrounding vehicles can obtain information, but it will occupy more communication channel resources and may cause channel congestion. The edge AI inference load refers to the proportion of computing resources allocated to edge computing nodes for running AI event detection and vehicle recognition, expressed as a percentage of CPU or GPU utilization. The higher the load, the more AI models can run simultaneously or the more complex the models, which may lead to increased processing latency.

[0027] Preferably, the constraint range refers to the bottom line that the decision variables cannot be breached when adjusted. It is set according to the physical limits and operational requirements of urban emergency management, including the minimum emergency lane release time, the maximum information underreporting rate, and the upper limit of communication bandwidth. Among them, the minimum emergency lane release time refers to the shortest duration during which the emergency lane must be kept open after it is opened to ensure that rescue vehicles have enough time to pass. The maximum information underreporting rate refers to the maximum allowable percentage of the number of traffic incidents that are not detected out of the total number of actual incidents. For example, the underreporting rate is stipulated not to exceed 5%. The upper limit of communication bandwidth refers to the maximum network bandwidth that can be used when data is transmitted between roadside equipment and the central system or between roadside equipment and vehicles. For example, the maximum uplink bandwidth of a 4G / 5G module is 100Mbps. When adjusting the video frame rate, radar point cloud update cycle, and V2X message frequency, the total amount of data generated cannot exceed this upper limit of bandwidth, otherwise it will lead to network congestion and data packet loss.

[0028] Furthermore, step S200 also includes accessing a data interaction interface, uploading event confidence, vehicle trajectory anomaly, and affected area range to construct a digital twin of the traffic event; performing dynamic simulation based on the digital twin of the traffic event, and feeding back the decision variables for optimizing the multi-objective optimization model.

[0029] Preferably, a data interaction interface is accessed to receive data in real time, and event confidence, vehicle trajectory anomaly, and affected area range are uploaded. Event confidence is the probability of identifying a certain event, and its value is usually between 0 and 1. For example, if the probability of judging "the current scene is a traffic accident" is 92%, then the uploaded event confidence is 0.92. Vehicle trajectory anomaly refers to the quantitative indicator of the deviation of a single vehicle's driving trajectory from the normal pattern, usually expressed as a deviation threshold or probability value. If a vehicle suddenly brakes, changes lanes continuously, or stops in the middle of the lane, the degree of anomaly in its trajectory is calculated. For example, an anomaly of 0.8 means that the vehicle's behavior deviates from the historical normal trajectory model by 80%. Affected area range refers to the geographical space range affected by the event. For example, the area affected by the event is "the section of XX Road from point M to point N". These real-time uploaded data are combined with static road map data and device location data to generate a digital twin of the traffic event that corresponds one-to-one with the real physical world and is synchronized in real time. Using the currently constructed digital twin as the initial state, a traffic flow simulation engine and an event propagation model are used to simulate the changing trends of the road network status over a very short period of time (e.g., 5 minutes, 10 minutes). This means that based on the current event, the model can quickly simulate possible future scenarios and obtain dynamic projection results, such as predicted queue lengths and congestion dissipation times. These results are then used as feedback signals to input into a multi-objective optimization model to optimize decision variables. For example, the weight of "congestion index at surrounding intersections" can be increased, while the priority of "emergency lane opening" can be reduced. This ensures that when a similar event occurs again, a better solution is generated that can both guarantee rescue and balance the surrounding road network.

[0030] Step S300: Simulate the road network response performance and event identification accuracy under different traffic event types, and generate emergency response plans for multimodal traffic events by combining the multi-objective optimization model.

[0031] Preferably, multiple preset emergency scenarios are used as simulation input conditions. For example, simulations include a two-vehicle rear-end collision occupying the leftmost lane, a vehicle fire requiring the closure of all lanes, severe flooding causing slow traffic, and a large event ending with a short-term surge in traffic. For each scenario, the actual traffic equipment parameters and road environment parameters are substituted to simulate the road network response efficiency and event recognition accuracy under different traffic event types. Among them, the road network response efficiency refers to the predicted changes in the road network operation status after different emergency strategies are adopted. For example, congestion dissipation time: how many minutes are expected for the road network to return to normal; maximum queue length: how many meters is expected to extend the congestion; capacity reduction rate: what percentage of the number of vehicles that can pass through the road segment per hour has decreased; and rescue vehicle arrival time: how long is expected for rescue vehicles to arrive at the scene after the emergency lane is opened. Event recognition accuracy refers to the probability of accurately detecting and classifying events, as well as false alarms and missed alarms, under current equipment parameters and environmental conditions. It outputs a prediction of detection performance based on the current sensor configuration. For example, under current visibility of 200 meters, the video algorithm's accuracy in identifying accidents is 95%. If the radar update cycle is extended from 100ms to 200ms, the accuracy of trajectory anomaly detection will decrease by 3 percentage points. Under the current scheme, it is estimated that there will be 2 false alarms (misclassifying sudden braking as an accident) out of every 100 detections. Then, the road network response performance and event recognition accuracy are input into a multi-objective optimization model for comprehensive scoring and selection. Each combination scheme is scored according to preset urban emergency management needs. Finally, the scheme with the highest score that meets all constraints is selected as the emergency response scheme for multimodal traffic events, which must at least include event ID, activation sequence, data acquisition equipment configuration parameters, and computing resource configuration data.

[0032] Step S400: Deploy edge computing nodes and configure traffic emergency control commands using the emergency response scheme.

[0033] Preferably, hardware devices with computing, storage, and network communication capabilities are installed and operated at locations close to the traffic incident site, such as intersections, roadside sections, and gantries. Edge computing nodes are deployed, such as intelligent traffic light controllers, intelligent cameras with AI computing chips, roadside intelligent base stations, or edge servers deployed in roadside cabinets. Then, the emergency response plan is parsed and translated into operation instructions for each edge computing node for specific hardware devices. For example, for a traffic light controller, an instruction is generated containing "Intersection ID=1001, Phase Number=4, Green Light Duration". For a roadside unit, an instruction containing "Message type = BSM, broadcast frequency = 10Hz, payload = accident location coordinates (x, y), broadcast power = 23dBm" is generated. The instruction is then sent through the network and written into the local storage or running memory of the corresponding edge computing node, putting it into a pending execution state. The edge computing node directly controls the traffic equipment through its physical output interface according to the configured instruction, such as traffic light switching, traffic information screen display, V2X broadcasting, etc.

[0034] Furthermore, the intelligent emergency response method based on multimodal traffic incident detection also includes deploying link test anchors at key communication nodes, including data interaction interfaces and V2X message forwarding nodes; when the message packet loss rate within the key communication node is detected to be higher than a threshold, communication relay enhancement is triggered.

[0035] Preferably, an active monitoring mechanism is deployed at key locations in the communication network to quantify communication quality in real time. When communication quality is detected to have fallen below a preset standard, enhanced communication reliability is automatically activated. Specifically, key communication nodes refer to specific network devices or logical interfaces in the emergency response data transmission path that perform data aggregation, protocol conversion, or broadcast forwarding functions. These may include data interaction interfaces or V2X message forwarding nodes, responsible for exchanging data with external platforms and broadcasting messages generated by roadside units to vehicle-mounted units or forwarding messages uploaded by vehicles to other roadside units. Deploying active monitoring mechanisms at key communication nodes... Link test anchors are set up to actively measure the quality of communication links. Test data packets are sent and received through the link test anchors, and the percentage of data packets sent within a certain period of time that did not receive acknowledgment packets is calculated to determine the message packet loss rate. The calculation formula is (total number of packets sent - total number of packets received) / total number of packets sent × 100%. This rate is compared with a threshold value, which is preset according to the service requirements of emergency communication. If the message packet loss rate is higher than the threshold value, it indicates a decline in communication quality, and communication relay enhancement is triggered, including power enhancement, path switching, relay forwarding, and encoding enhancement, to ensure that emergency instructions and data can be delivered.

[0036] Furthermore, the intelligent emergency response method based on multimodal traffic incident detection also includes: matching and determining communication assurance priorities based on a traffic incident modal knowledge base and the dynamic inference results of the traffic incident digital twin; and dynamically adjusting the strength and timing of communication relay enhancements according to the communication assurance priorities.

[0037] Preferably, a modal knowledge base containing communication requirements for different event types is established. Combined with predictions of future event evolution, the importance ranking of different communication nodes or data streams is calculated, and limited communication enhancement resources are allocated differentially based on this ranking. Specifically, the traffic event modal knowledge base is used to store the communication requirement characteristics corresponding to different types of historical traffic events. For example, for vehicle fires, key communication requirements are high-resolution video transmission (requires >10Mbps bandwidth), V2X instructions from rescue vehicles (requires <50ms latency), and broadcasts from surrounding vehicles (requires high reliability); for minor rear-end collisions, key communication requirements are low-resolution video confirmation (requires 2Mbps bandwidth) and broadcasts of the accident location (allowing for second-level latency); for road collapses, key communication requirements are multi-angle video transmission, radar point cloud data, and emergency broadcasts from all surrounding vehicles (extremely high priority). Then, based on the dynamic simulation results of the traffic incident digital twin, dynamic adjustments are made to match and determine the communication guarantee priority for each communication node. For example, P0 is the video stream from camera 1 at the accident scene and the two-way command stream between the rescue vehicle and the center; P1 is the radar data stream at the accident scene and the update command from the traffic information screen at the upstream intersection; P2 is the routine traffic data statistics stream from the surrounding unaffected areas and the upload of remote maintenance logs. The strength and timing of communication relay enhancements are then dynamically adjusted, including automatically configuring different levels of communication enhancement measures according to priority, and automatically determining the order and duration of enhancement measures activation based on priority and the simulation results of the event evolution.

[0038] Furthermore, the intelligent emergency response method based on multimodal traffic incident detection also includes constructing a traffic incident modal knowledge base, which stores emergency response time requirements and multimodal perception resource allocation strategies under the annotation of traffic accident mode, vehicle spontaneous combustion mode, and road collapse mode.

[0039] Preferably, the characteristics, processing requirements, and resource needs of different types of traffic incidents are stored in a standardized format to construct a structured traffic incident modal knowledge base. The traffic incident types include traffic accident modes, vehicle spontaneous combustion modes, and road collapse modes. The traffic incident modal knowledge base stores emergency response time requirements and multimodal perception resource allocation strategies under the annotations of traffic accident modes, vehicle spontaneous combustion modes, and road collapse modes. Specifically, traffic accident modes are typical accident scenarios such as collisions, scrapes, and rear-end collisions between motor vehicles. The mode identifiers are rear-end collision accidents, side collision accidents, etc. The feature tags are the number of vehicles involved, the lane position occupied (leftmost / rightmost / middle), whether pedestrians are involved, and whether dangerous goods transport vehicles are involved. The emergency response time requirements include a delay of less than 3 seconds for issuing the first priority instruction and a time of less than 10 seconds for the emergency lane to be fully opened. The multimodal perception resource allocation strategy includes a video sensor resource ratio of 60%, a radar sensor resource ratio of 30%, and a geomagnetic / coil sensor resource ratio of 10%. Two of the three cameras are set to high frame rate mode and aimed at the accident point.

[0040] Preferably, the vehicle spontaneous combustion mode is for scenarios where a motor vehicle catches fire due to a malfunction or accident. The mode is identified as spontaneous combustion, and the mode labels include the speed of fire spread, whether thick smoke is produced, and whether it is necessary to evacuate surrounding personnel. The emergency response timeliness requirements are that the broadcast delay of the vehicle evacuation instruction within 200 meters is less than 1 second, the interface trigger time for notifying the fire department is less than 2 seconds, and the phase lock-up time of all surrounding traffic lights is less than 5 seconds. The multimodal perception resource allocation strategy includes 80% video sensor resources, 15% meteorological sensor resources, and 5% V2X message resources, and all surrounding cameras with thermal imaging capabilities are mobilized to the vicinity of the fire point.

[0041] Preferably, the road collapse mode is for scenarios where potholes, subsidence, or collapse suddenly appear on the road surface. The mode is identified as collapse, and the mode labels include collapse area, collapse depth, whether it affects the roadbed, and whether there is a risk of secondary collapse. The emergency response timeliness requirements are: lane closure command delay in the collapse area is less than 2 seconds; all traffic information screens within 2 kilometers upstream publish detour information within 15 seconds; and data push interface trigger time of the navigation platform is less than 20 seconds. The multimodal perception resource allocation strategy includes 60% radar sensor resources, 35% video sensor resources, and 5% LiDAR resources, reducing the video analysis load in irrelevant directions and concentrating computing resources on analyzing radar point cloud data in the collapse area.

[0042] In the above text, refer to Figure 1 This paper describes in detail an intelligent emergency response method based on multimodal traffic incident detection according to embodiments of the present invention. Next, reference will be made to... Figure 2 This invention describes an intelligent emergency response system based on multimodal traffic incident detection according to an embodiment of the present invention.

[0043] The intelligent emergency response system based on multimodal traffic incident detection according to embodiments of the present invention addresses the technical problems in existing technologies, such as lagging emergency response mechanisms, rigid contingency plans, low resource scheduling efficiency, and lack of dynamic simulation and verification of road network performance. It achieves the technical effects of improving the real-time performance, accuracy, and dynamic adaptability of traffic incident emergency response, optimizing edge computing resource utilization, and enhancing the overall operational efficiency of the road network. Figure 2 As shown, the intelligent emergency response system based on multimodal traffic incident detection includes: a data acquisition module 10, a dynamic adjustment module 20, a scheme generation module 30, and an instruction configuration module 40.

[0044] The data acquisition module 10 is used to acquire technical parameters of traffic equipment and road environment parameters; the dynamic adjustment module 20 is used to configure a multi-objective optimization model that meets the needs of urban emergency management based on the technical parameters of traffic equipment and road environment parameters, and dynamically adjust the activation sequence of emergency plans, the frequency of multimodal perception data collection, and the edge computing resource allocation strategy; the scheme generation module 30 is used to simulate the road network response efficiency and event recognition accuracy under different traffic event types, and generate emergency response schemes under multimodal traffic events in combination with the multi-objective optimization model; the instruction configuration module 40 is used to deploy edge computing nodes and configure traffic emergency control instructions using the emergency response scheme.

[0045] The specific configuration of the dynamic adjustment module 20 will be described in detail below. The dynamic adjustment module 20 further includes: setting a constraint range by using emergency response start-up delay, video analysis frame rate, radar point cloud update cycle, V2X message broadcast frequency, and edge AI inference load as decision variables; the constraint range includes minimum emergency lane release time, maximum information false alarm rate, and upper limit of communication bandwidth.

[0046] The specific configuration of the dynamic adjustment module 20 will be described in detail below. The dynamic adjustment module 20 further includes: accessing a data interaction interface to upload event confidence levels, vehicle trajectory anomalies, and affected area ranges to construct a digital twin of the traffic event; performing dynamic simulations based on the digital twin of the traffic event; and feeding back the decision variables for optimizing the multi-objective optimization model.

[0047] The specific configuration of the instruction configuration module 40 will be described in detail below. The instruction configuration module 40 further includes: the traffic emergency control instructions include traffic light emergency phase switching instructions, emergency lane opening instructions, and surrounding intersection diversion instructions; based on the edge computing node, a synchronization mechanism is configured on the data interaction interface, the synchronization mechanism being used to correct the transmission delay of the traffic emergency control instructions.

[0048] The following section will continue to describe in detail the specific configuration of the intelligent emergency response system based on multimodal traffic incident detection. It further includes: deploying link test anchor points at key communication nodes, including data interaction interfaces and V2X message forwarding nodes; and triggering communication relay enhancement when the message packet loss rate within the key communication nodes exceeds a threshold.

[0049] The following section will continue to describe in detail the specific configuration of the intelligent emergency response system based on multimodal traffic incident detection. It further includes: determining communication assurance priorities based on a traffic incident modal knowledge base and the dynamic inference results of the traffic incident digital twin; and dynamically adjusting the strength and timing of communication relay enhancements according to the communication assurance priorities.

[0050] The specific configuration of the strategy generation module 40 will be described in detail below. The strategy generation module 40 further includes: constructing a traffic event modal knowledge base, which stores emergency response time requirements and multimodal perception resource allocation strategies under the annotation of traffic accident mode, vehicle spontaneous combustion mode, and road collapse mode.

[0051] The intelligent emergency response system based on multimodal traffic incident detection provided in this embodiment of the invention can execute the intelligent emergency response method based on multimodal traffic incident detection provided in this embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0052] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An intelligent emergency response method based on multimodal traffic incident detection, characterized in that, The method includes: Obtain technical parameters of traffic equipment and road environment parameters; Based on the technical parameters of the traffic equipment and the road environment parameters, a multi-objective optimization model that meets the needs of urban emergency management is configured to dynamically adjust the activation sequence of emergency plans, the frequency of multimodal sensing data collection, and the edge computing resource allocation strategy. The road network response efficiency and event identification accuracy under different traffic event types are simulated, and the multi-objective optimization model is combined to generate emergency response plans for multimodal traffic events. Deploy edge computing nodes and configure traffic emergency control commands using the aforementioned emergency response scheme.

2. The intelligent emergency response method based on multimodal traffic incident detection as described in claim 1, characterized in that, Based on the aforementioned traffic equipment technical parameters and road environment parameters, a multi-objective optimization model that meets the needs of urban emergency management is configured. The method further includes: The emergency response activation delay, video analysis frame rate, radar point cloud update cycle, V2X message broadcast frequency, and edge AI inference load are used as decision variables, and constraints are set. The constraints include the minimum emergency lane release time, the maximum information miss rate, and the upper limit of communication bandwidth.

3. The intelligent emergency response method based on multimodal traffic incident detection as described in claim 2, characterized in that, The method includes: Connect to the data interaction interface, upload event confidence level, vehicle trajectory anomaly level and affected area range to build a digital twin of traffic events; Dynamic simulations are performed based on the digital twin of the traffic incident, and the decision variables of the multi-objective optimization model are fed back to optimize the model.

4. The intelligent emergency response method based on multimodal traffic incident detection as described in claim 3, characterized in that, The method for configuring traffic emergency control commands using the aforementioned emergency response scheme includes: The traffic emergency control commands include emergency phase switching commands for traffic lights, commands to open emergency lanes, and commands to divert traffic from surrounding intersections. Based on the edge computing node, a synchronization mechanism is configured on the data interaction interface. The synchronization mechanism is used to correct the transmission delay of the traffic emergency control command.

5. The intelligent emergency response method based on multimodal traffic incident detection as described in claim 1, characterized in that, The method includes: Link test anchor points are deployed at key communication nodes, including data interaction interfaces and V2X message forwarding nodes. When the message packet loss rate within the critical communication node is detected to be higher than the threshold, communication relay enhancement is triggered.

6. The intelligent emergency response method based on multimodal traffic incident detection as described in claim 5, characterized in that, The method includes: Based on the traffic incident modal knowledge base and combined with the dynamic inference results of the traffic incident digital twin, the priority of communication protection is matched and determined. The strength and timing of communication relay enhancements are dynamically adjusted based on the communication guarantee priority.

7. The intelligent emergency response method based on multimodal traffic incident detection as described in claim 6, characterized in that, The method includes: A traffic event modal knowledge base is constructed, which stores emergency response time requirements and multimodal perception resource allocation strategies under the annotation of traffic accident mode, vehicle spontaneous combustion mode, and road collapse mode.

8. An intelligent emergency response system based on multimodal traffic incident detection, characterized in that, The system is used to implement the intelligent emergency response method based on multimodal traffic incident detection as described in any one of claims 1 to 7, and the system comprises: The data acquisition module is used to acquire technical parameters of traffic equipment and road environment parameters; The dynamic adjustment module is used to configure a multi-objective optimization model that meets the needs of urban emergency management based on the technical parameters of the traffic equipment and the road environment parameters, and to dynamically adjust the activation sequence of the emergency plan, the frequency of multimodal sensing data collection, and the edge computing resource allocation strategy. The scheme generation module is used to simulate the road network response efficiency and event identification accuracy under different traffic event types, and generate emergency response schemes under multimodal traffic events in combination with the multi-objective optimization model. The instruction configuration module is used to deploy edge computing nodes and configure traffic emergency control instructions using the aforementioned emergency response scheme.

Images