Starburst technology based traffic control system and method

Abstract

The application discloses a traffic control system and method based on star flash technology, and relates to the technical field of intelligent traffic control. The system comprises an edge computing gateway and a star flash sensing node group and a traffic signal lamp group arranged at a traffic intersection. The star flash sensing node group is used for collecting radar sensing data of people and vehicles at the traffic intersection and transmitting the radar sensing data to the edge computing gateway. The edge computing gateway is used for taking a preset lightweight congestion level judgment strategy as an operation model, dynamically calculating the green light release duration of people and vehicles based on the radar sensing data, generating a release control instruction corresponding to the green light release duration, and controlling the corresponding traffic signal lamp group based on the release control instruction. The application is used for solving the problems that the existing traffic control scheme has poor scene adaptation, high transformation cost, and large intersection complex scheduling scheme deployment to small intersection causes waste of computing power resources, and cannot realize dynamic no-countdown self-adaptive timing of small intersection.

Description

Technical Field

[0001] This invention relates to the field of intelligent traffic control technology, and in particular to a traffic control system and method based on star-flash technology. Background Technology

[0002] Urban intersections can be broadly categorized into large arterial road intersections with independent left-turn phases and multiple lanes, and small, simple intersections (crossroads / T-junctions) on secondary arterial roads, side roads, and in residential areas without complex turning phases. Large arterial road intersections have complex traffic flow patterns, requiring multi-phase control equipment for flow management. However, directly deploying such costly, complex multi-phase control equipment and algorithms at small, simple intersections would result in a significant waste of hardware investment and computing resources. Small intersections only need to distinguish between pedestrians and straight-going vehicles, without requiring complex turning logic.

[0003] Currently, regardless of whether the intersection is large or small, traffic lights mostly adopt a control mode with a fixed phase cycle and a fixed digital countdown. The supporting vehicle and pedestrian flow detection solutions mainly use cameras, geomagnetic sensors, and infrared sensors, along with traditional Bluetooth and WiFi wireless modules to complete data transmission. The existing system has poor adaptability and many inherent defects: 1) Poor environmental adaptability of sensing equipment: Cameras are easily affected by strong light, rain, snow, fog, haze, and obstruction; geomagnetic sensors can only identify motor vehicles and cannot collect pedestrian data; infrared devices can only roughly determine the presence of targets and cannot quantify queue length and crowd density, and their all-weather stable detection capability is insufficient; 2) Insufficient performance of traditional wireless transmission: The electromagnetic environment at intersections is complex, and conventional radio frequency wireless solutions have weak anti-interference capabilities and high data transmission delays, which easily lead to the loss of sensing data and delayed issuance of release control commands; 3) Rigid timing mechanism: The red and green light release times are fixed in advance and cannot be adaptively adjusted according to the real-time traffic flow during morning and evening peak hours, off-peak hours, and nighttime, frequently resulting in long green lights for empty lanes and insufficient pedestrian crossing time, leading to low utilization of road traffic resources.

[0004] Existing perception and control solutions on the market are either designed for large main roads and equipped with complex multi-phase steering scheduling logic, making them extremely cost-effective for deployment at small intersections; or they adopt a uniform fixed timing mode, which cannot be dynamically adjusted according to the real-time pedestrian and vehicle traffic flow at small intersections. Summary of the Invention

[0005] The purpose of this invention is to provide a traffic control system and method based on star-flash technology, which can overcome the shortcomings of existing traffic control schemes, such as poor scenario adaptability, high transformation cost, waste of computing resources caused by deploying complex scheduling schemes for large intersections to small intersections, and inability to achieve dynamic, countdown-free adaptive timing for small intersections.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a traffic control system based on starlight technology, the system comprising: an edge computing gateway deployed at a traffic intersection, and a group of starlight sensing nodes and a group of traffic lights that are communicatively connected to the edge computing gateway; The StarFlash sensing node group is used to collect radar sensing data of people and vehicles at traffic intersections and transmit it to the edge computing gateway; The edge computing gateway is used to dynamically calculate the green light duration for pedestrians and vehicles based on radar perception data, using a preset lightweight congestion level judgment strategy as the computing model. It generates a release control command corresponding to the green light duration and controls the corresponding traffic signal light group based on the release control command.

[0007] Optionally, using a preset lightweight congestion level judgment strategy as the calculation model, and based on radar perception data, the green light duration is dynamically calculated, including: The waiting objects are selected based on radar perception data, and the values ​​of multiple congestion indicators corresponding to the waiting objects are statistically analyzed. Based on the values ​​of multiple congestion indicators, a weighted fusion method is used to determine the current congestion level. Retrieve the green light duration range corresponding to the current congestion level, and calculate the actual green light duration using linear interpolation. Verify the actual green light duration; if the actual green light duration is less than the preset minimum safe green light duration, then use the preset minimum safe green light duration as the actual green light duration.

[0008] Optionally, based on the values ​​of multiple congestion indicators, a weighted fusion method is used to determine the current congestion level, specifically including: The values ​​of each congestion indicator are matched with a preset congestion indicator segmentation table to obtain the score of each indicator. The preset weighting coefficients of each indicator are retrieved, and the scores of each indicator are weighted and summed to calculate the comprehensive congestion index. The overall congestion index is matched with a preset overall index level comparison table to determine the current congestion level.

[0009] Optionally, the star-flash sensing node group includes pedestrian star-flash sensing nodes, which are deployed in the pedestrian walkway area of ​​traffic intersections.

[0010] Optionally, the star-flash sensing node group includes vehicle star-flash sensing nodes; the vehicle star-flash sensing nodes are deployed in the motor vehicle lane area of ​​the traffic intersection.

[0011] Optionally, the radar sensing data includes pedestrian coordinates and pedestrian speed within the pedestrian electronic fence; The waiting groups were identified based on radar sensing data, and the values ​​of several congestion indicators corresponding to these groups were statistically analyzed, including: Pedestrians whose speed is less than a preset slowing threshold are identified as the waiting pedestrian group; Based on pedestrian coordinates, overlapping identification points in the waiting pedestrian group whose coordinate spacing is less than the preset pedestrian clustering spacing are clustered to remove duplicates, and the number of valid pedestrians is counted and the coordinates of all valid pedestrians are output. The vertical distance from each valid pedestrian to the pedestrian stop line is calculated based on the coordinates of all valid pedestrians, and the maximum value among all vertical distances is taken as the size of the pedestrian queue.

[0012] Optionally, the radar sensing data includes vehicle coordinates and vehicle radial velocity within a preset sliding statistical window; The waiting groups were identified based on radar sensing data, and the values ​​of several congestion indicators corresponding to these groups were statistically analyzed, including: Vehicles with radial speeds less than a preset speed threshold are identified as waiting vehicles, and the coordinates of all waiting vehicles are extracted. Based on the coordinates of all waiting vehicles, calculate the straight-line distance from the farthest waiting vehicle to the vehicle stop line to obtain the vehicle queue length; The number of valid vehicles within the sliding statistics window is counted to obtain the real-time traffic flow. Combined with the single-lane saturation capacity, the lane saturation is calculated. Extract the radial velocities of all vehicles within the sliding statistical window and take their arithmetic mean to obtain the lane average speed.

[0013] Optionally, the traffic signal group includes two sets of traffic lights; the two sets of traffic lights are pedestrian traffic lights and vehicle traffic lights.

[0014] Optionally, the traffic signal group includes a set of shared traffic lights for both pedestrians and vehicles; When traffic lights are shared, pedestrians and vehicles with higher congestion levels are given priority for green light duration when determining the congestion level.

[0015] Compared with existing technologies, the traffic control system based on starlight technology provided by this invention relies on an integrated starlight sensing node group to collect radar sensing data of pedestrians and vehicles. It eliminates the need for multiple separate detection devices such as cameras, geomagnetic sensors, and infrared sensors, simplifying the hardware layout of intersections and reducing the investment in supporting wiring, equipment procurement, and subsequent operation and maintenance. This effectively reduces the overall cost of intelligent transformation of small intersections and overcomes the high cost of traditional solutions. Simultaneously, the system uses a lightweight congestion level judgment strategy as the local computing model of the edge gateway. Based on real-time radar sensing data, it dynamically generates green light durations and corresponding release control commands adapted to pedestrian and vehicle traffic flow. It eliminates the need for preset fixed phase cycles and countdown logic, adapting to the lightweight computing power deployment requirements of small, simple intersections in side roads and communities. This solves the technical shortcomings of traditional fixed timing solutions, such as poor scenario adaptability and inability to dynamically adjust timing as needed, achieving countdown-free adaptive intelligent traffic signal control at small intersections.

[0016] Secondly, the present invention also provides a traffic control method based on star-flash technology, the method comprising: The edge computing gateway acquires radar perception data of people and vehicles; the radar perception data is collected by the star-flash perception node group. Using a preset lightweight congestion level judgment strategy as the calculation model, and based on radar perception data, the green light duration for pedestrians and vehicles is dynamically calculated and a release control command corresponding to the green light duration is generated. The corresponding traffic light group is controlled based on the release control command. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 One of the schematic diagrams of a traffic control system based on star-flash technology is provided as an embodiment of the present invention; Figure 2 A second schematic diagram of a traffic control system based on star-flash technology, provided as an embodiment of the present invention; Figure 3 A third schematic diagram of a traffic control system based on star-flash technology, provided as an embodiment of the present invention; Figure 4 This is a flowchart illustrating a traffic control method based on star-flash technology, provided as an embodiment of the present invention.

[0018] Figure label: 10-Edge computing gateway; 20-Starlight sensing node group; 210-Pedestrian starlight sensing node; 220-Vehicle starlight sensing node; 30-Traffic signal light group; 301-Pedestrian traffic light; 302-Vehicle traffic light; 303-Shared traffic light; 40-Cloud management platform. Detailed Implementation

[0019] To facilitate a clear description of the technical solutions in the embodiments of the present invention, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first threshold and the second threshold are merely used to distinguish different thresholds and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" are not necessarily different.

[0020] It should be noted that in this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0021] In this invention, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between the associated objects, indicating that three relationships can exist.

[0022] like Figure 1 As shown, this embodiment of the invention provides a traffic control system based on starlight technology. The system includes: an edge computing gateway 10 deployed at a traffic intersection, and a starlight sensing node group 20 and a traffic signal light group 30 that are communicatively connected to the edge computing gateway. The Star Flash Sensing Node Group 20 is used to collect radar sensing data of people and vehicles at traffic intersections and transmit it to the edge computing gateway 10; The edge computing gateway 10 is used to dynamically calculate the green light duration for pedestrians and vehicles based on radar perception data, using a preset lightweight congestion level judgment strategy as the computing model, generating a release control command corresponding to the green light duration, and controlling the corresponding traffic signal light group 30 based on the release control command.

[0023] Specifically, such as Figure 1 As shown, the star-flash sensing node group 20 includes a pedestrian star-flash sensing node 210 and a vehicle star-flash sensing node 220.

[0024] Both the pedestrian XingShan sensing node 210 and the vehicle XingShan sensing node 220 are equipped with XingShan Near Link (a new generation of short-range wireless communication standard) SLP (Sensing & Link Perception, a sensing and positioning protocol) dual-mode integrated single-chip. The entire sensing and communication functions are integrated into the same dedicated SoC (System on Chip). This single chip integrates four core components: XingShan's self-developed millimeter-wave radar module, multi-antenna array RF transceiver components, high-precision clock synchronization unit, and low-power XingShan wireless communication baseband. The independent functions of each component are as follows: The millimeter-wave radar module is responsible for transmitting and receiving radio frequency radar echo signals, completing the active detection of both human and vehicle targets. It is the core of the star-flash sensing node for realizing distance measurement, speed measurement, and feature recognition of human and vehicle targets.

[0025] The multi-antenna array RF transceiver module, in conjunction with the millimeter-wave radar module, completes AoA (Angle of Arrival) direction finding signal acquisition, while also carrying out star-flash wireless data transmission and reception, enabling distributed self-organizing network multi-hop transmission between intersection nodes.

[0026] The high-precision clock synchronization unit provides a unified clock reference at the microsecond level, enabling timing alignment of all star-flash sensing nodes at the intersection and ensuring the synchronous fusion of multi-node collaborative sensing data.

[0027] The low-power star-flash wireless communication baseband parses and encapsulates Near Link protocol data packets to complete the uplink transmission of sensing data and the downlink transmission of gateway control commands, and has industrial-grade electromagnetic interference resistance capabilities.

[0028] The chip achieves time-division synchronous operation of communication and radar perception by sharing spectrum resources and reusing software and hardware, without the need for additional independent radar, cameras, or infrared detection peripherals. This node relies on radio frequency sensing capabilities such as ToF (Time of Flight), AoA, Doppler frequency shift velocity measurement, and RCS (Radar Cross Section) feature analysis to output a complete set of raw quantitative sensing data in parallel. This includes multiple types of raw data such as the target's straight-line distance d, two-dimensional plane coordinates (x, y), target radial velocity, radar cross section (RCS), and target continuous detection duration. Simultaneously, the star-flash sensing node has a built-in filtering unit for radar clutter filtering, human / vehicle target identification, and anti-aliasing purification of multi-target radar ghost images. It only performs basic cleaning at the radio frequency raw signal level, outputting a set of basic target coordinates after removing false alarm interference. This unit only performs low-level signal preprocessing and does not perform business indicator calculations such as spatial clustering of waiting crowds or people counting. Subsequent coordinate-based clustering deduplication and effective pedestrian count logic are completed by the edge computing gateway and do not participate in congestion indicator numerical calculations.

[0029] Vehicle star-flash sensing nodes are deployed in the motor vehicle lane area of ​​traffic intersections. The number of vehicle star-flash sensing nodes and pedestrian star-flash sensing nodes can be set according to actual conditions, and this invention does not impose any restrictions.

[0030] The beneficial effects of this embodiment are: 1) Reduced cost of intersection reconstruction and full life-cycle operation and maintenance: The StarFlash sensing node integrates radar detection and wireless communication functions, which can replace multiple separate devices such as traditional cameras, geomagnetic sensors, infrared sensors, and independent wireless transmission devices, reducing the hardware and labor investment in peripheral procurement, wiring construction, and later maintenance, and significantly reducing the cost of intelligent intersection reconstruction; 2) Adaptable to lightweight deployment of small intersections: The system in this embodiment adopts a lightweight congestion judgment calculation model, which has low requirements for edge gateway computing power and does not require the use of complex multi-phase scheduling hardware at the main road level. It can be deployed at low cost for small branch roads and simple community intersections, without the waste of idle hardware and computing power resources, overcoming the problem of traditional solutions using a one-size-fits-all approach for large and small intersections and poor adaptability to small intersections; 3) Realized dynamic adaptive timing of pedestrian and vehicle traffic flow at intersections without countdown, improving... Road traffic resource utilization: In this embodiment, the system relies on the star-flash sensing node to collect real-time radar sensing data of people and vehicles. The lightweight model of the edge gateway calculates the green light duration adapted to the current traffic flow in real time. There is no preset fixed signal cycle or fixed countdown. The release time can be flexibly adjusted according to the number of waiting people and vehicles in real time, avoiding the waste of resources such as green light when there are no vehicles or insufficient waiting time for pedestrians, and improving the traffic efficiency of intersections; 4) All-weather stable full-domain synchronous perception of people and vehicles, ensuring the data foundation for dynamic timing: Star-flash radar perception is not affected by strong light, rain, snow, fog, haze, or obstruction. It can collect data of both pedestrians and vehicles simultaneously. Compared with traditional single motor vehicle detection equipment, the perception coverage is more comprehensive, providing complete and reliable real-time traffic data support for the dynamic timing calculation of the edge gateway, and ensuring the stable operation of the adaptive timing strategy.

[0031] In an exemplary embodiment, a preset lightweight congestion level judgment strategy is used as the computational model. Based on radar perception data, the green light duration is dynamically calculated, including: Step 1: Filter out waiting objects based on radar perception data and count the values ​​of multiple congestion indicators corresponding to the waiting objects; Step 2: Based on the values ​​of multiple congestion indicators, a weighted fusion method is used to determine the current congestion level; For example, based on the values ​​of multiple congestion indicators, a weighted fusion method is used to determine the current congestion level, specifically including: (1) Match the values ​​of each congestion index with the preset congestion index segmentation table to obtain the score of each index; retrieve the preset weight coefficients of each index, sum the scores of each index by weight, and calculate the comprehensive congestion index. (2) Match the comprehensive congestion index with the preset comprehensive index level comparison table to determine the current congestion level.

[0032] Step 3: Retrieve the green light duration range corresponding to the current congestion level, and calculate the actual green light duration using linear interpolation; verify the actual green light duration. If the actual green light duration is less than the preset minimum safe green light duration, then use the preset minimum safe green light duration as the actual green light duration.

[0033] For example, the actual green light duration is calculated using linear interpolation, specifically including: Formula used: (1); The actual green light duration was calculated. This refers to the actual green light duration. This represents the lower limit of the green light duration range corresponding to the current congestion level. This represents the upper limit of the green light duration range corresponding to the current congestion level; This represents the minimum overall congestion index corresponding to the current congestion level. This represents the maximum value of the comprehensive congestion index corresponding to the current congestion level. The comprehensive congestion index is calculated by weighting the current target (pedestrians / motor vehicles); where formula (1) conforms to the interpolation boundary lock rule: 1) if the comprehensive congestion index is lower than the lower limit of the index of this level, i.e. Take the shortest duration of the interval directly: 2) If the overall congestion index is higher than the upper limit of this level, i.e. Take the longest duration in the interval directly: 3) If the interpolation calculates If the actual green light duration is less than the preset minimum safe time for pedestrians (e.g., 8 seconds), the actual green light duration will be forcibly adjusted to the preset minimum safe time for pedestrians.

[0034] The beneficial effects of this embodiment are as follows: 1) The use of multi-indicator weighted fusion unified judgment logic ensures that the congestion level judgment result is objective and controllable: Multiple congestion indicators corresponding to the waiting objects are assigned scores respectively, and then a comprehensive congestion index is obtained by weighted summation through preset weights. The congestion level is determined based on the comprehensive index, avoiding the one-sidedness brought about by single indicator judgment. The standardized weighted calculation achieves quantitative and objective congestion classification, reducing subjective judgment bias; 2) The layered and progressive calculation process is logically clear and can be verified step by step: The overall process is divided into multiple progressive steps: indicator scoring → weighted calculation of comprehensive index → ​​matching congestion level → linear interpolation to calculate green light duration → safe duration verification. Each step outputs intermediate observable values, which facilitates program logic verification and fault location, and the operation link is structured; 3) Linear interpolation achieves fine-grained timing within the level range, and the duration adjustment granularity is finer: Not each congestion level corresponds to only a single fixed green light duration, but the precise release duration is calculated by linear interpolation based on the range. Within the same congestion level, the green light duration can be finely adjusted according to real-time changes in indicators, and the timing granularity is finer. Detailed, without completely fixed durations within each gear level; 4) Dual duration constraint checks are set to ensure timing compliance: on the one hand, linear interpolation naturally limits the duration to fall within the corresponding range of the current gear level; on the other hand, an additional minimum safe green light duration check is added as a fallback; the two constraints jointly limit the final release time, matching the current congestion level range without causing unreasonable green light durations below the safety threshold, balancing traffic flow adaptation and traffic safety; 5) The calculation relies only on table lookup, weighted summation, and linear arithmetic operations, resulting in lightweight and low computing power consumption: throughout the entire process It only uses basic mathematical operations such as segmented lookup table matching, score weighting, and linear interpolation, without complex iterations or nonlinear solutions, and the lightweight operation logic can be executed in real time on embedded edge devices; 6) The indicator scoring, weight, level, and duration interval are decoupled and the parameters are independently adjustable: the indicator segmented lookup table, weight coefficient, comprehensive index level lookup table, and green light duration interval for each level are mutually independent configuration items. Modifying any type of parameter does not require changing the overall hierarchical calculation steps, which facilitates the separate configuration of adaptation parameters for different intersection scenarios.

[0035] In an exemplary embodiment, the radar sensing data includes the coordinates and speed of pedestrians within the pedestrian electronic fence; the waiting object is a group of waiting pedestrians; Waiting objects are identified based on radar perception data, and the values ​​of multiple congestion indicators corresponding to the waiting objects are statistically analyzed. For pedestrians, waiting pedestrian groups are identified based on pedestrian radar perception data, and the values ​​of multiple pedestrian congestion indicators corresponding to the waiting pedestrian groups are statistically analyzed. Furthermore, for pedestrians, waiting pedestrian groups are identified based on pedestrian radar perception data, and the values ​​of several pedestrian congestion indicators corresponding to these groups are statistically analyzed, including: Pedestrians whose speed is less than a preset slowing threshold (e.g., 0.3 m / s) are identified as the waiting pedestrian group; Based on pedestrian coordinates, overlapping identification points in the waiting pedestrian group with coordinate spacing less than the preset pedestrian clustering spacing (e.g., 0.8m) are clustered for deduplication, and the number of valid pedestrians is obtained. Output the coordinates of all valid pedestrians; The vertical distance from each valid pedestrian to the pedestrian stop line is calculated based on the coordinates of all valid pedestrians. The maximum vertical distance among all pedestrian distances is taken as the pedestrian queue size. The number of effective pedestrians and the size of pedestrian queues are used as indicators of pedestrian congestion.

[0036] After obtaining the values ​​of multiple pedestrian congestion indicators, the scores of two pedestrian congestion indicators, namely the effective number of pedestrians and the size of pedestrian queues, are obtained according to the preset pedestrian congestion indicator segmentation comparison table. The preset weight coefficients of the two pedestrian congestion indicators, namely the number of effective pedestrians and the size of the pedestrian queue, are retrieved. The scores of the two pedestrian congestion indicators, namely the number of effective pedestrians and the size of the pedestrian queue, are weighted and summed to calculate the comprehensive pedestrian congestion index. Match the pedestrian congestion index with a preset pedestrian congestion level comparison table to determine the current pedestrian congestion level; Finally, the actual green light duration for pedestrians is calculated using formula (1).

[0037] It should be noted that, in one optional implementation, after determining the actual green light duration, if a large number of people and vehicles are added to the waiting period during the passage, the actual green light duration can be appropriately extended; if the traffic flow in the passage direction dissipates quickly and the minimum safe duration is met, the traffic lights can be switched in advance.

[0038] The specific traffic control process for pedestrians is illustrated in Specific Scenario Example 1: First, the waiting pedestrian group is screened within the pedestrian electronic fence, and the number of valid pedestrians is counted. Pedestrian queue size : The coordinates of the four corners of the pedestrian electronic fence are as follows: , , , The preset slow-moving threshold is 0.3m / s, and the preset pedestrian clustering distance is 0.8m.

[0039] Within the electronic fence of the pedestrian crossing area at the intersection, the star-flash sensing node collected data on 13 pedestrians. Four pedestrians had a speed > 0.3 m / s and were identified as passing pedestrians, thus being removed. The remaining nine pedestrians had a speed ≤ 0.3 m / s and were considered waiting pedestrians. Five pairs of coordinates among these nine waiting pedestrians were less than 0.8 m apart. After clustering and deduplication, the final number of valid pedestrians was determined. Then, calculate the vertical distance from all 14 valid pedestrians to the pedestrian stop line. The maximum value is 12m, which represents the size of the pedestrian queue. .

[0040] Secondly, based on the indicator segmentation comparison table (Table 1), obtain the individual scores for the two indicators: Table 1: Segmented Scoring Comparison Table for Pedestrian Dual Indicators As can be seen from Table 1, Falling within the interval The corresponding score is ; Falling within the interval Corresponding score .

[0041] Next, the weights are retrieved to calculate the overall pedestrian congestion index. Specifically, the weight for retrieving the number of valid pedestrians. Weighting of pedestrian queue size Substitute the values ​​of the two weights mentioned above into the formula for calculating the pedestrian congestion index: (2); Calculated .

[0042] Next, match the comprehensive index comparison table (Table 2) to determine the pedestrian congestion level: Table 2: Pedestrian Comprehensive Congestion Index and Level Comparison Table This calculation yielded ,satisfy The current pedestrian congestion level is determined to be level 3 (medium concentration), corresponding to the basic green light duration range [20s, 32s].

[0043] Finally, adopting the method of , , , , Substituting into formula (1), the actual green light duration for pedestrians is calculated. .

[0044] The beneficial effects of this embodiment are as follows: 1) By defining a dedicated electronic fence for the pedestrian crossing to limit the statistical area, and then using a preset slow-moving threshold to filter waiting crowds, only slow-moving targets within the fence are counted. Pedestrians moving quickly across or traversing the pedestrian crossing are not included in the calculation, thus avoiding irrelevant targets interfering with the congestion index value from the source and ensuring that pedestrian timing only matches the actual waiting demand for crossing the street; 2) The radar in the pedestrian star-flash sensing node will generate multiple overlapping coordinate identification points for the same group of pedestrians. The scheme sets a preset clustering spacing to merge adjacent targets, eliminating the problem of inflated numbers caused by radar ghosting, so that the effective number of pedestrians can truly reflect the actual scale of waiting pedestrians, preventing excessive redundancy in green light time due to duplicate counting, and balancing traffic efficiency; 3) Selecting two pedestrian congestion indicators: the effective number of pedestrians can reflect the total scale of the waiting crowd, and the pedestrian queue size can reflect the depth of the queue extending backward. Relying solely on the number of people cannot distinguish between "a small number of pedestrians close to the stop line" and "a large number of pedestrians queuing a long distance backward". The system offers two differentiated congestion scenarios: "staffing" and "waiting." The dual-indicator combination can fully cover the characteristics of pedestrian plane clustering, ensuring that timing judgments are not one-sided. 4) Dedicated weights are set for pedestrians. For example, the total number of people waiting can be used as the core weight for congestion judgment, which aligns with the core needs of pedestrians. When many people gather on the pedestrian crossing, the comprehensive congestion index increases rapidly, and a longer green light is automatically allocated to solve the problem of insufficient green light time for pedestrians gathering at small intersections. 5) Pedestrian indicators do not require iterative calculations such as time-series sliding windows and flow difference. They only complete clustering and distance calculation based on single-frame radar coordinates and speed. The calculation steps are simple, adaptable to lightweight deployment of edge gateways, and will not consume too much hardware computing resources.

[0045] In an exemplary embodiment, the radar sensing data includes vehicle coordinates and vehicle radial speed within a preset sliding statistical window; Waiting objects are identified based on radar perception data, and the values ​​of multiple congestion indicators corresponding to the waiting objects are statistically analyzed. For motor vehicles, waiting vehicles are identified based on vehicle radar perception data, and the values ​​of multiple vehicle congestion indicators corresponding to the waiting vehicles are statistically analyzed. Furthermore, for motor vehicles, waiting vehicles are identified based on vehicle radar perception data, and the values ​​of multiple vehicle congestion indicators corresponding to the waiting vehicles are statistically analyzed, specifically including: (1) Identify vehicles whose radial speed is less than a preset vehicle speed threshold (e.g., 0.5 m / s) as waiting vehicles and extract the coordinates of all waiting vehicles; (2) Based on the coordinates of all waiting vehicles, calculate the straight-line distance from the farthest waiting vehicle to the vehicle stop line to obtain the vehicle queue length. Among them, when there are no vehicles ; (3) Count the number of valid vehicles in the sliding statistics window to obtain the real-time traffic flow, and calculate the lane saturation by combining the single lane saturation capacity (e.g., 18 vehicles / 10s). In the specific implementation of step (3), the formula is adopted: (3); Calculate real-time traffic flow; among which, Real-time traffic flow; The number of valid vehicles identified within a preset sliding statistics window; This is a preset sliding statistics window; Formula used: (4); Calculate lane saturation; where, Lane saturation; The single-lane saturation capacity can be preset to a specific value.

[0046] (4) Extract the radial speed of all vehicles in the sliding statistical window and take the arithmetic mean to obtain the average speed of the lane; use the vehicle queue length, lane saturation and average speed of the lane as multiple vehicle congestion indicators.

[0047] After obtaining the values ​​of multiple vehicle congestion indicators, the scores of three vehicle congestion indicators—vehicle queue length, lane saturation, and average lane speed—are obtained based on the preset vehicle congestion indicator segmentation comparison table. The weighting coefficients of the three vehicle congestion indicators—vehicle queue length, lane saturation, and average lane speed—are retrieved, and the scores of vehicle queue length, lane saturation, and average lane speed are weighted and summed to calculate the comprehensive vehicle congestion index. Match the vehicle congestion index with a preset vehicle congestion level table to determine the current vehicle congestion level. Finally, the actual green light duration for vehicles is calculated using formula (1).

[0048] The specific traffic control process for vehicles is illustrated in Example 2 of a specific scenario: First, waiting vehicles are filtered based on vehicle radar perception data within a preset sliding statistical window, and the queue length is calculated. Lane saturation Average speed per lane Three indicators of vehicle congestion: Preset sliding statistics window The vehicle speed threshold for determining a stationary vehicle is 0.5 m / s, and the single-lane saturation traffic volume is... Vehicles / 10s.

[0049] The vehicle detection nodes on the motor vehicle lane identified 27 vehicle targets within a 10-second sliding statistical window; these were further filtered by vehicle radial velocity. There were 16 vehicles waiting at a standstill; the rest were vehicles that were moving normally and were directly removed. The coordinates of all 16 waiting vehicles were extracted, and the straight-line distance was calculated using the lane stop line as a reference. The farthest waiting vehicle was 32 meters from the stop line, which is the length of the vehicle queue. .

[0050] Total number of valid vehicles in this window Vehicles, substitute into formula (3) to calculate real-time traffic flow: ; Then substitute the traffic flow into formula (4) to calculate lane saturation: ; Extract the radial velocity values ​​of all 27 vehicles within the sliding statistical window and calculate their arithmetic mean to obtain the lane average speed. .

[0051] The final values ​​for the three vehicle congestion indicators were determined: , , .

[0052] Secondly, based on the vehicle indicator segmentation comparison table (Table 3), the individual scores for the three vehicle congestion indicators are obtained: Table 3: Segmented Scoring Comparison Table for Three Motor Vehicle Indicators Based on the item-by-item matching score in Table 3: queue length Falling in the range Corresponding score Lane saturation Falling within the interval Corresponding score Average vehicle speed Falling within the interval Corresponding score .

[0053] Next, the weighting of vehicle indicators is retrieved to calculate the overall vehicle congestion index. : Specifically, retrieve the queue length weight. Lane saturation weight Average vehicle speed weight Substitute into the formula for calculating the comprehensive congestion index of motor vehicles: (5); Substitute the numerical values ​​into the calculation: .

[0054] Next, match the comprehensive vehicle index table (Table 4) to determine the level of vehicle congestion: Table 4: Comparison of Comprehensive Motor Vehicle Congestion Index and Gear Levels This calculation yielded ,satisfy The current vehicle congestion level is determined to be level 3 (medium concentration), corresponding to the basic green light duration range [25s, 40s].

[0055] Finally, , , , , Substituting into the linear interpolation formula (1), the actual green light duration for vehicles is calculated as follows: .

[0056] The beneficial effects of this embodiment are as follows: 1) The vehicle uses a unified sliding window to cyclically collect targets and periodically outputs traffic flow and average speed, which can continuously capture real-time changes in traffic flow. Compared with single-frame static detection, it can stably obtain the increase in vehicles entering the lane per unit time, providing standardized time-series data for lane saturation calculation; 2) It adopts three independent vehicle congestion indicators: vehicle queue length, lane saturation, and average lane speed. Vehicle queue length represents the current static backlog in the lane, lane saturation represents the vehicle inflow load per unit time, and average lane speed distinguishes between slow-moving and completely stalled traffic flows. A single indicator can only reflect local congestion characteristics, but the combination of the three can simultaneously identify various conditions such as short-term traffic influx, long queues, and completely deadlocked queues; 3) It does not directly use the original effective vehicle count for scoring. By converting traffic flow into a dimensionless saturation ratio, it unifies the evaluation criteria for lanes of different widths and carrying capacities. In scenarios with multiple flow directions and multiple lane specifications running in parallel, the degree of congestion is reduced. 4) Stationary waiting vehicles are screened separately using a speed threshold of, for example, 0.5 m / s. The queue length is calculated based solely on the coordinates of stationary targets, and vehicles traveling at normal constant speeds are not included in the queue statistics. This method can accurately separate dynamic traffic flow, avoid passing vehicles from lengthening the queue calculation distance, and ensure that the vehicle queue length only represents the actual size of the waiting queue. 5) Differentiated weights are set for vehicle indicators, with the queue length, which is most intuitive for drivers and intersections, set as the core weight. In the scenario of long, stationary queues in all lanes, the differentiation of low-weight average vehicle speed decreases, but high-weight queue length can still dominate the judgment result, ensuring that the congestion index is not distorted. 6) The average speed of all vehicles in the collection window is taken, rather than only the speed of queued vehicles. This can capture the pre-congestion state of slow movement at the front of the queue and continuous influx at the back. The congestion score can be increased before vehicles have completely stopped queuing, thereby extending the green light time in advance and suppressing the aggravation of congestion.

[0057] In an exemplary embodiment, the edge computing gateway further includes a forward warning unit, which is used to trigger a congestion forward warning when the increase in traffic flow exceeds a preset traffic flow threshold or the increase in vehicle queue length exceeds a preset length.

[0058] In practice, the growth rate of traffic flow The difference in traffic flow between two adjacent sliding statistical windows: see formula (6): (6); in, This is the current sliding statistics window; This is the previous sliding statistics window; The actual length of the sliding statistics window, for example, 10 seconds.

[0059] Increase in vehicle queue length The difference in queue length between two adjacent sliding statistical windows is given in formula (7): (7); For example, when two consecutive windows or When a traffic congestion warning is triggered, the green light duration will only be taken within the range corresponding to the current congestion level. It cannot jump to a higher congestion level. The release time will be pre-allocated according to the upper limit of the range to suppress traffic accumulation in advance.

[0060] The beneficial effects of this embodiment are as follows: 1) Calculating the comprehensive congestion index solely based on vehicle queue length, lane saturation, and average lane speed within a single window is a reactive, post-event judgment that can only allocate green lights to congestion that has already formed. However, the increase in traffic flow and the increase in vehicle queue length are captured by using a time-series difference calculation through a 10-second sliding window. Even if the current comprehensive congestion index is still at a medium level, if the traffic flow and queues increase rapidly for two consecutive cycles, the green light limit for this level can be reached in advance, diverting traffic before large-scale queues form and preventing congestion from worsening and spreading to intersections. 2) Pedestrian gatherings are mostly instantaneous and one-time clusters, without the temporal evolution characteristics of continuous vehicle inflow and continuous queue lengthening, thus eliminating the need for time-series difference calculations. Vehicles exhibit a dynamic evolution pattern of continuous entry and progressively lengthening queues, requiring only the addition of a speed-up judgment logic for vehicles, without superimposing time-series iteration calculations on pedestrian flow. While achieving congestion prediction, it will not unnecessarily increase the computational load of pedestrian indicators, maintain the overall lightweight computing power architecture, and adapt to low-cost edge gateways for small intersections; 3) At the same time, it sets a threshold for incremental traffic flow and an incremental threshold for incremental vehicle queue length, which can be triggered when either is met, covering two types of congestion precursors: "sudden increase in traffic flow" and "rapid lengthening of queues"; with the superimposed verification mechanism of two consecutive sliding statistical windows, the instantaneous traffic flow fluctuations in a single window (such as a single large vehicle or instantaneous radar clutter) will not misjudge the warning, filter random noise, and the accuracy of the pre-intervention strategy is higher, and it will not occupy the opposite passage time for no reason; 4) The warning trigger is strictly limited to the green light duration range corresponding to the current vehicle congestion level, and it is prohibited to call higher-level durations across levels, and it will not directly jump to the duration range of higher congestion levels. It achieves both advance capacity expansion and strict adherence to the four-level congestion classification and linear interpolation framework, without disrupting the original segmented scoring and weighted index judgment logic. The parameter system is unified and controllable, eliminating the need to reconstruct the basic timing model, and reducing the difficulty of subsequent parameter debugging and intersection adaptation; 5) The growth rate calculation fully reuses the original time-series stored data of vehicle indicators, without the need to add radar sampling frequency or new hardware acquisition units. Only the difference arithmetic operation is added in the edge gateway, and the congestion prediction capability can be obtained with zero hardware modification cost. There is no additional power consumption or transmission overhead at the star flash sensing node; 6) If the queue is extended to the level of severe congestion before the green light is extended, vehicles are very likely to cut in, weave, and rush through the yellow light; the advance warning extends the release time in advance, smoothly digests the influx of traffic, controls the growth rate of the queue length, maintains the continuous passage of traffic, reduces low-speed weaving, rapid acceleration, and emergency braking behavior at intersections, and simultaneously improves traffic efficiency and intersection traffic safety.

[0061] Regarding traffic signal light systems, in Specific Implementation Method 1, as follows: Figure 1 As shown, the traffic signal group includes two sets of traffic lights; the two sets of traffic lights are pedestrian traffic light 301 and vehicle traffic light 302.

[0062] In the second specific implementation method, such as Figure 2As shown, the traffic signal group includes a set of shared traffic lights 303 for both pedestrians and vehicles. When the traffic signal group is a shared traffic light group, when the congestion level is determined, pedestrians and vehicles with higher congestion levels are given priority in being allocated green light duration.

[0063] Among them, pedestrian traffic lights, vehicle traffic lights and shared traffic lights have completely eliminated countdown screens, and only retain the three-color light prompts of red, yellow and green lights; a fixed yellow light buffer time (e.g. 3 seconds) is uniformly set, and after the green light is dynamically adjusted, it switches to yellow light, and then jumps to the opposite green light, forming a complete switching logic without countdown signals.

[0064] In an exemplary embodiment, see Figure 3 The system may also include a cloud management platform 40; the cloud management platform 40 is communicatively connected to the edge computing gateway 10, and the cloud management platform 40 is used for remote monitoring and parameter configuration.

[0065] like Figure 4 As shown, this embodiment of the invention also provides a traffic control method based on star-flash technology, the method comprising: Step 400: The edge computing gateway acquires radar perception data of people and vehicles; the radar perception data is collected by the star-flash perception node group; In practice, radar perception data includes pedestrian radar perception data and vehicle radar perception data.

[0066] Step 410: Using the preset lightweight congestion level judgment strategy as the calculation model, based on radar perception data, dynamically calculate the green light release time for people and vehicles and generate release control instructions corresponding to the green light release time. The preset lightweight congestion level judgment strategy is divided into two categories: lightweight pedestrian congestion level judgment strategy and lightweight vehicle congestion level judgment strategy. The overall calculation framework of the two strategies is unified, and both follow the standardized process of "target selection → indicator statistics → segmented scoring → weighted calculation of comprehensive congestion index → ​​matching level → linear interpolation calculation of green light duration". There are four differences, and the remaining calculation rules, interpolation formulas and minimum safe duration constraint logic are completely consistent. (1) The original perception data and the waiting object screening rules are different: pedestrians rely on the pedestrian electronic fence on the sidewalk and the slow walking threshold to screen waiting pedestrians, and the coordinates are clustered to remove duplicates; vehicles rely on the sliding statistical window and the vehicle speed threshold to screen stationary waiting vehicles, and the traffic flow data is continuously collected in time series. (2) The congestion evaluation index system and index weights are configured independently: pedestrians use two indicators, effective pedestrian number and pedestrian queue size, with exclusive weights; vehicles use three indicators, vehicle queue length, lane saturation and average lane speed, with exclusive weights. (3) The segmented scoring comparison table, the comprehensive congestion index level, and the green light duration interval are independent of each other, and pedestrians and vehicles are set with differentiated four-level thresholds and safe green light lower limits; (4) The vehicle strategy adds a congestion warning logic: predict the traffic flow outbreak by the traffic flow growth rate and queue length growth rate of adjacent sliding statistical windows, and the pedestrian growth rate warning judgment process has no time sequence.

[0067] Step 420: Control the corresponding traffic signal light group based on the release control command.

[0068] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the appended claims in carrying out the claimed invention. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality.

[0069] Although the invention has been described in conjunction with specific features and embodiments, it is apparent that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely illustrative descriptions of the invention as defined by the appended claims and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A traffic control system based on star-flash technology, characterized in that, The system includes: an edge computing gateway deployed at traffic intersections, and a group of star-flash sensing nodes and a group of traffic lights that are communicatively connected to the edge computing gateway; The star-flash sensing node group is used to collect radar sensing data of people and vehicles at the traffic intersection and transmit it to the edge computing gateway; The edge computing gateway is used to dynamically calculate the green light duration for pedestrians and vehicles based on the radar perception data, using a preset lightweight congestion level judgment strategy as the computing model, generating a release control command corresponding to the green light duration, and controlling the corresponding traffic signal light group based on the release control command.

2. The traffic control system based on star-flash technology according to claim 1, characterized in that, The method of using a preset lightweight congestion level judgment strategy as a calculation model and dynamically calculating the green light duration based on the radar perception data includes: Based on the radar sensing data, waiting objects are selected, and the values ​​of multiple congestion indicators corresponding to the waiting objects are statistically analyzed. Based on the values ​​of multiple congestion indicators, a weighted fusion method is used to determine the current congestion level. The green light duration range corresponding to the current congestion level is retrieved, and the actual green light duration is calculated using linear interpolation. The actual green light duration is verified. If the actual green light duration is less than the preset minimum safe green light duration, the preset minimum safe green light duration is used as the actual green light duration.

3. The traffic control system based on star-flash technology according to claim 2, characterized in that, The current congestion level is determined by using a weighted fusion method based on the values ​​of multiple congestion indicators, specifically including: The values ​​of each of the multiple congestion indicators are matched with a preset congestion indicator segmentation table to obtain the score of each indicator. The preset weighting coefficients of each indicator are retrieved, and the scores of each indicator are weighted and summed to calculate the comprehensive congestion index. The current congestion level is determined by matching the comprehensive congestion index with a preset comprehensive index level comparison table.

4. The traffic control system based on star-flash technology according to claim 3, characterized in that, The star-flash sensing node group includes pedestrian star-flash sensing nodes, which are deployed in the pedestrian walkway area of ​​the traffic intersection.

5. The traffic control system based on star-flash technology according to claim 4, characterized in that, The star-flash sensing node group includes vehicle star-flash sensing nodes; the vehicle star-flash sensing nodes are deployed in the motor vehicle lane area of ​​the traffic intersection.

6. The traffic control system based on star-flash technology according to claim 2, characterized in that, The radar sensing data includes pedestrian coordinates and speeds within the pedestrian electronic fence. The process involves filtering waiting objects based on the radar sensing data and statistically analyzing the values ​​of multiple congestion indicators corresponding to the waiting objects, including: Pedestrians whose speed is less than a preset slowing threshold are identified as a waiting pedestrian group; Based on the pedestrian coordinates, the overlapping identification points in the waiting pedestrian group whose coordinate spacing is less than the preset pedestrian clustering spacing are clustered and deduplicated, the number of valid pedestrians is counted and all valid pedestrian coordinates are output. The vertical distance from each valid pedestrian to the pedestrian stop line is calculated based on the coordinates of all valid pedestrians, and the maximum value among all vertical distances is taken as the size of the pedestrian queue.

7. The traffic control system based on star-flash technology according to claim 2, characterized in that, The radar sensing data includes vehicle coordinates and vehicle radial speed within a preset sliding statistical window; The process involves filtering waiting objects based on the radar sensing data and statistically analyzing the values ​​of multiple congestion indicators corresponding to the waiting objects, including: Vehicles with radial speeds less than a preset speed threshold are identified as waiting vehicles, and the coordinates of all waiting vehicles are extracted. Based on the coordinates of all waiting vehicles, calculate the straight-line distance from the farthest waiting vehicle to the vehicle stop line to obtain the vehicle queue length; The number of valid vehicles within the sliding statistics window is counted to obtain the real-time traffic flow. Combined with the single-lane saturation capacity, the lane saturation is calculated. The radial speeds of all vehicles within the sliding statistical window are extracted and their arithmetic averages are taken to obtain the average speed of the lane.

8. The traffic control system based on star-flash technology according to claim 1, characterized in that, The traffic signal light group includes two sets of traffic lights; the two sets of traffic lights are pedestrian traffic lights and vehicle traffic lights.

9. The traffic control system based on star-flash technology according to claim 8, characterized in that, The traffic signal light group includes a set of shared traffic lights for both pedestrians and vehicles; When the traffic light group is a shared traffic light, when the congestion level is determined, pedestrians and vehicles with higher congestion levels are given priority in being allocated green light duration.

10. A traffic control method based on star-flash technology, characterized in that the method include: The edge computing gateway acquires radar perception data of people and vehicles; The radar sensing data is collected by the star flash sensing node group; Using a preset lightweight congestion level judgment strategy as the calculation model, based on the radar perception data, the green light release time for people and vehicles is dynamically calculated and a release control command corresponding to the green light release time is generated. The corresponding traffic light group is controlled based on the release control command.

Images