LED lamp intelligent guiding control method and system and LED lamp

Abstract

The invention relates to the technical field of illumination control, in particular to an LED lamp intelligent guiding control method and system and an LED lamp. The method comprises the following steps: acquiring environment sensing data of a target area; identifying a motion intention of each moving target in the target area based on the target area environment perception data to obtain moving target motion intention data; recognizing obstacles in the target area according to the moving target motion intention data, and generating an area dynamic obstacle semantic graph; generating a candidate guiding path set according to the regional dynamic obstacle semantic graph; generating an initial guiding path diagram based on the candidate guiding path set; and detecting a conflict path of the initial guiding path diagram, and performing path avoidance processing on the conflict path to obtain an optimal guiding path diagram. According to the invention, the blank of the LED guide lamp system in dynamic obstacle collaborative avoidance strategies is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of lighting control, and particularly to an intelligent guiding control method, system and LED lamp for LED lamps. Background Art

[0002] Most of the existing LED guiding lamp systems realize their functions based on fixed programming or simple infrared sensing technology. These systems can meet the basic vehicle guiding needs in daily use, but their limitations gradually emerge when facing some complex dynamic working conditions. For example, in public areas with large traffic flow (such as parking lots, transportation hubs, large commercial centers), when there are sudden changes in parking space occupancy, temporary road closures or sudden changes in traffic flow, the existing systems cannot adjust the guiding path of the guiding lamp in time, resulting in chaos when vehicles are looking for parking spaces or leaving parking spaces.

[0003] In addition, in emergency situations, such as fire drills and emergency evacuation scenarios, the defects of the existing systems are further amplified. During a fire drill, a group of fire-fighting equipment needs to enter or leave an underground parking lot, building or traffic artery quickly and orderly to simulate a real fire rescue scenario. However, since the existing LED guiding lamp systems cannot identify a moving group of fire-fighting equipment, the escape guiding path conflicts with the advancing route of the fire-fighting equipment. This conflict not only interferes with the normal advancement of the fire-fighting equipment, but also easily forces emergency vehicles to brake suddenly, thus delaying the rescue time. Similarly, in an emergency evacuation scenario, the existing systems cannot dynamically adjust the evacuation path according to the real-time personnel distribution and dynamic obstacle situation, resulting in low evacuation efficiency. This phenomenon fully exposes the gap in the dynamic obstacle cooperative avoidance strategy of traditional LED guiding lamp systems. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide an intelligent guiding control method, system and LED lamp for LED lamps to solve at least one of the above technical problems.

[0005] To achieve the above object, an intelligent guiding control method for an LED lamp includes the following steps: Step S1: Obtain environmental perception data of a target area; Step S2: Based on the environmental perception data of the target area, identify the movement intentions of each moving target in the target area to obtain moving target movement intention data; identify obstacles in the target area according to the moving target movement intention data, and generate a regional dynamic obstacle semantic map; Step S3: Generate a candidate guiding path set according to the regional dynamic obstacle semantic map; generate an initial guiding path map based on the candidate guiding path set; detect conflicting paths in the initial guiding path map, and perform path avoidance processing on the conflicting paths to obtain an optimal guiding path map; Step S4: Perform instruction partition collaborative control on the LED lamp system according to the optimal guiding path map to obtain a synchronous LED control instruction set; configure the dynamic effect parameters of the LED lamps according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table; Step S5: Perform a lighting effect rendering guiding control operation on the LED lamp system according to the LED dynamic display parameter table.

[0006] By obtaining the environmental perception data of the target area and identifying the movement intentions of moving targets and dynamic obstacles, the present invention can generate an optimal guiding path map in real time and collaboratively control the LED lamp system, thus effectively solving the limitations of the existing system under dynamic working conditions. It can quickly respond to emergencies such as parking space changes, avoid chaos when vehicles are searching for or leaving parking spaces, improve the traffic fluency of the parking lot and eliminate potential safety hazards. In emergency scenarios such as fire drills, the present invention can identify dynamic obstacles such as fire equipment groups, and through path avoidance processing, ensure that its travel route does not conflict with the escape indication path, avoid interfering with the normal progress of fire equipment, prevent emergency vehicles from making emergency brakes, thus ensuring the rescue efficiency and preventing secondary accidents caused by defects in the guiding light system, significantly improving the intelligence level and safety of the LED guiding light system in the parking lot.

[0007] Preferably, the present invention provides an LED lamp intelligent guiding control system for executing the above-mentioned LED lamp intelligent guiding control method. The LED lamp intelligent guiding control system includes: An environmental perception module for obtaining environmental perception data of the target area; An intention and obstacle recognition module for identifying the movement intentions of each moving target in the target area based on the environmental perception data of the target area to obtain moving target movement intention data; identifying obstacles in the target area according to the moving target movement intention data to generate a regional dynamic obstacle semantic map; A guiding path planning module for generating a candidate guiding path set according to the regional dynamic obstacle semantic map; generating an initial guiding path map based on the candidate guiding path set; detecting conflicting paths in the initial guiding path map and performing path avoidance processing on the conflicting paths to obtain an optimal guiding path map; An LED parameter configuration module for performing instruction partition collaborative control on the LED lamp system according to the optimal guiding path map to obtain a synchronous LED control instruction set; configuring the dynamic effect parameters of the LED lamps according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table; An LED guiding control module for performing a lighting effect rendering guiding control operation on the LED lamp system according to the LED dynamic display parameter table.

[0008] In the present invention, the environmental perception module can obtain comprehensive perception data in the parking lot in real time, providing a basis for subsequent precise control and ensuring the system's rapid response to the dynamic situation in the parking lot. Through the intention and obstacle recognition module, the movement intention of the vehicle and dynamic obstacles can be accurately identified, effectively avoiding potential risks brought by the unpredictability of vehicle behavior and improving the safety of the parking lot. Through the guiding path planning module, an optimal guiding path can be generated based on the dynamic obstacle semantic map, solving the static limitations of traditional systems in path planning. Especially in complex dynamic scenarios, path conflicts can be effectively avoided, ensuring the orderly passage of vehicles. Through the LED parameter configuration module, refined zoning collaborative control of the LED lighting system is achieved, dynamically adjusting the lighting effect parameters according to the optimal path, enhancing the intuitiveness and accuracy of the guidance. The LED guiding control module realizes the efficient guidance of vehicles through lighting effect rendering, enhancing the visibility and guiding effect of the system. In summary, the present invention not only improves the operation efficiency and safety of the parking lot, but also provides reliable guiding support for emergency scenarios, reducing the accident risk caused by improper guidance.

[0009] Preferably, the present invention also provides an LED lamp for performing the above-mentioned intelligent guiding control method of the LED lamp. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Other features, objects, and advantages of the present invention will become more apparent by reading the following detailed description with reference to the accompanying drawings: Figure 1 The flowchart of the steps of the intelligent guiding control method of the LED lamp according to an embodiment is shown.

[0011] Figure 2 The detailed flowchart of the steps of step S1 according to an embodiment is shown.

[0012] Figure 3 The detailed flowchart of the steps of step S4 according to an embodiment is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those skilled in the art within the scope of the present invention without creative efforts belong to the scope of protection of the present invention.

[0014] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus the repeated description thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0015] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0016] To achieve the above object, please refer to Figures 1 to 3 , the present invention provides an intelligent guiding control method for an LED lamp, including the following steps: Step S1: Obtain the environmental perception data of the target area; Step S2: Based on the environmental perception data of the target area, identify the movement intentions of each moving target in the target area to obtain moving target movement intention data; identify the obstacles in the target area according to the moving target movement intention data, and generate a regional dynamic obstacle semantic map; Step S3: Generate a candidate guiding path set according to the regional dynamic obstacle semantic map; generate an initial guiding path map based on the candidate guiding path set; detect the conflicting paths of the initial guiding path map, and perform path avoidance processing on the conflicting paths to obtain an optimal guiding path map; Step S4: Perform instruction partition collaborative control on the LED lamp system according to the optimal guiding path map to obtain a synchronous LED control instruction set; configure the dynamic effect parameters of the LED lamp according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table; Step S5: Perform a lighting effect rendering guiding control operation on the LED lamp system according to the LED dynamic display parameter table.

[0017] In this embodiment, the embodiments of the present application will be described by taking the scenario of a parking lot as an example. For example, the environmental perception data of the target area is the environmental perception data of the parking lot; the target area is the parking lot, and the moving target is a vehicle. The vehicle thermal map data is collected by an infrared sensor array (such as FLIR ThermoVision A40) installed in the parking lot, the distance to obstacles is measured by an ultrasonic sensor array (such as HC-SR04), the vehicle movement speed is obtained by combining a microwave radar (such as Infineon BGT24M16C), and the visual image data of the target area is collected by a camera. These data are integrated into the environmental perception data of the target area through timestamp calibration and three-dimensional space mapping. Then, the movement intention of the moving target is recognized based on the environmental perception data of the target area. Using image processing and machine learning libraries in MATLAB or Python (such as OpenCV and scikit-learn), the vehicle contour, morphological features, and light signal features are extracted. Through matching and classification using a preset feature vector library, the vehicle type and status are determined, and the movement intention of the vehicle, such as going straight, turning left, or turning right, is predicted by combining the vehicle's historical behavior data. The movement intention data of the moving target is converted into a structured semantic description using semantic mapping rules, and a multi-level semantic layer is constructed by combining the situational semantic data of the moving target, including single-vehicle behavior, fleet behavior, and area behavior. Through emergency state feature extraction and moving target priority classification, a regional dynamic obstacle semantic map is generated. Based on the dynamic obstacle semantic map, a candidate guiding path set is generated using a path planning algorithm (such as Dijkstra's algorithm). By combining the basic topology map of the target area and the digital twin model, the passing cost of each passable grid is calculated, a path planning basic cost map is generated, and multiple candidate paths are searched. By calculating the path congestion index and path safety score, each candidate path is comprehensively evaluated, and the optimal path is selected to form an initial guiding path map. A path scheduling strategy is constructed using the graph database Neo4j, conflict points are identified and conflict identification data is generated, and a path conflict solution is generated through a decision tree algorithm to finally obtain the optimal guiding path map. The floor plan of the target area is meshed using AutoCAD and Python scripts, and the path points are mapped to the grid cells to generate a path-grid correspondence table. Through clustering analysis, blocks with similar guiding functions are divided to generate LED light control function blocks. The actual LED light position distribution map is obtained, the LED light control function blocks are mapped to the actual LED light positions, the control partition to which each LED light belongs is determined, and the control priority is assigned according to the functional characteristics to generate a regional LED control partition map. A synchronous LED control instruction set is generated according to the regional LED control partition map, and the dynamic effect parameters of the LED lights are configured.Using MATLAB or Python combined with a custom control algorithm, generate dynamic effect parameters such as brightness, color, and blinking frequency according to the functional characteristics of the control zones (such as guidance, warning, and no-entry), and obtain the LED dynamic display parameter table. Finally, perform lighting effect rendering and guidance control operations on the LED lighting system according to the LED dynamic display parameter table. Send control instructions to the LED lights through an intelligent control system (such as PLC or Arduino) to achieve dynamic display and guidance control of the lights.

[0018] Preferably, step S1 includes the following steps: Step S11: Collect moving target thermal map data through an infrared sensor array and collect target area obstacle distance data through an ultrasonic sensor array; Specifically, an infrared sensor array and an ultrasonic sensor array can be installed at the entrance and key passage positions of the parking lot. The infrared sensor array uses thermal imaging technology and can detect the thermal radiation signal of the vehicle, thereby generating moving target thermal map data. For example, use the ThermoVision A40 infrared thermal imager, which can capture the thermal distribution of vehicles in the parking lot in real time. At the same time, the ultrasonic sensor array is used to measure the distance data of obstacles in the parking lot. Select the HC-SR04 ultrasonic sensor, which calculates the distance to the obstacle by emitting an ultrasonic signal and receiving the reflected signal. These sensors are evenly distributed on the walls and columns of the parking lot, and one is installed every certain distance (such as 3 meters). When a vehicle enters the parking lot, the infrared sensor array captures the thermal signal of the vehicle and generates thermal map data, while the ultrasonic sensor array measures the distance between the vehicle and surrounding obstacles, and the two work together.

[0019] Step S12: Use a microwave radar to collect the moving speed of the moving target and collect visual image data of the target area through a camera to form a visual image sequence of the target area, where the microwave radar sampling rate ranges from 10 Hz to 100 Hz and the camera frame rate ranges from 15 fps to 60 fps; Specifically, a microwave radar and a camera can be installed at key positions in the parking lot. The microwave radar uses a 24 GHz band Doppler radar, such as the BGT24M16C microwave radar chip. The radar can accurately measure the moving speed of the vehicle by emitting a microwave signal and receiving the reflected signal. The radar is installed on the ceiling of the parking lot, covering the lane area below, and can monitor the driving speed of the vehicle in real time. At the same time, the camera is installed on the wall of the parking lot at a certain angle, and one is installed every about 5 meters to form a continuous visual image sequence. The microwave radar and the camera work simultaneously. The radar records the moving speed data of the vehicle, and the camera collects the visual image data in the parking lot.

[0020] Step S13: Perform timestamp calibration on the moving target heat map data and the target area obstacle distance data to obtain the moving target-obstacle time-aligned data; Specifically, a GPS synchronization module can be used, such as the Garmin GPS 18x-5Hz module, which can provide accurate time signals for synchronizing the timestamps of all sensors. During system initialization, connect the GPS synchronization module to the infrared sensor array and the ultrasonic sensor array to ensure that the time bases of all devices are consistent. When the sensors start collecting data, the GPS synchronization module will timestamp each data point. Read these timestamped data files through data processing software (such as data processing libraries in MATLAB or Python), and align the vehicle heat map data and the obstacle distance data according to the timestamps. For example, if the infrared sensor captures the heat signal of the vehicle at a certain moment, and the ultrasonic sensor measures the distance to the obstacle near the same timestamp, the software will associate these two pieces of data to generate the moving target-obstacle time-aligned data.

[0021] Step S14: Construct a three-dimensional space coordinate system; based on the three-dimensional space coordinate system, integrate the moving target motion speed, the moving target-obstacle time-aligned data, and the target area visual image sequence to obtain the target area environmental perception data.

[0022] Specifically, a laser scanner (such as the FARO Focus3D X 330) can be used to perform three-dimensional modeling of the parking lot to obtain the spatial dimensions and layout information of the parking lot. The laser scanner calculates the three-dimensional coordinates of each point in the space by emitting laser beams and measuring the reflection time, thereby generating a three-dimensional point cloud model of the parking lot. Import the moving target-obstacle time-aligned data and the target area visual image sequence into three-dimensional modeling software (such as Rhinoceros). In the software, use the three-dimensional point cloud model as the basic framework, and map the vehicle heat map data and the obstacle distance data to the corresponding three-dimensional space positions according to the timestamps and sensor position information. At the same time, project each frame of the target area visual image sequence into the three-dimensional space according to the position and angle information of the camera. Through a data fusion algorithm (such as the Kalman filter algorithm), integrate these data from different sources to generate the target area environmental perception data containing vehicle position, speed, thermal characteristics, obstacle distance, and visual image information. These data are presented in a three-dimensional visualization manner, which can intuitively display the real-time dynamic situation in the parking lot.

[0023] The present invention can comprehensively and real-time sense the vehicle and obstacle information in the parking lot by respectively collecting vehicle thermal map data and obstacle distance data through an infrared sensor array and an ultrasonic sensor array, and combining a microwave radar and a camera to obtain the vehicle movement speed and visual image sequence. The synchronization and accuracy of multi-source data are further ensured through timestamp calibration and three-dimensional space mapping.

[0024] Preferably, in step S2, identifying the movement intentions of each moving target in the target area based on the target area environmental perception data specifically includes: Extracting the regional contours of each moving target in the target area from the target area environmental perception data to obtain a moving target candidate area map; Specifically, in the target area environmental perception data, the OpenCV image processing library can be used to extract the vehicle regional contours. Input the visual image sequence in the target area environmental perception data into OpenCV, and preprocess the image through grayscale conversion and edge detection algorithms (such as Canny edge detection). Use the contour finding function cv2.findContours to detect all the contours in the image. Set an area threshold (for example, 1000 pixel squares), filter out the contours with too small area, so as to screen out the potential regional contours belonging to the vehicle. Draw these contours on a blank image to generate a moving target candidate area map. For example, if there is an SUV and a sedan in the parking lot, their contours can be extracted respectively through the above method and clearly shown in the moving target candidate area map.

[0025] Extracting the morphological features of the moving targets in the moving target candidate area map to obtain a moving target type feature vector, and matching and classifying the moving target type feature vector with a preset moving target type feature vector library to obtain a moving target type recognition result; Specifically, the moving target candidate area map can be imported into MATLAB, and the regionprops function can be used to calculate the morphological features of each vehicle contour, such as aspect ratio, area, and perimeter. Taking the aspect ratio as an example, the calculation formula is the ratio of the width to the height of the vehicle contour. Combine these feature values into a feature vector. For example, for an SUV, its feature vector is [aspect ratio = 1.5, area = 2000 pixel squares, perimeter = 180 pixels]. The preset moving target type feature vector library stores the standard feature vectors of different vehicle models. Through similarity measurement methods such as Euclidean distance, the extracted feature vector is matched and classified with the vectors in the library. If the distance between the extracted feature vector and the standard feature vector of the SUV is the smallest, it is determined that the vehicle is an SUV, and a moving target type recognition result is obtained.

[0026] Identifying the movement state of the moving target based on the target area environmental perception data to form moving target state feature data; Specifically, frame-by-frame analysis can be performed on the visual image sequence of the target area. By setting a brightness threshold (for example, an area with a pixel value greater than 200 is considered a bright spot), the position of the bright spot of the vehicle lamp can be detected. The position changes of each bright spot in consecutive frames are statistically analyzed. If the position of the bright spot remains basically unchanged in several consecutive frames and the brightness changes regularly, it is determined that the vehicle lamp is flashing. For example, if it is detected that a certain bright spot flashes once every 3 frames in 10 consecutive frames of images, its flashing mode is recorded as "periodic flashing, with a period of 3 frames". The distribution positions and flashing modes of all vehicle lamp bright spots are summarized to form the state feature data of the moving target.

[0027] Determine the target type-state recognition result based on the moving target type recognition result and the state feature data of the moving target; Specifically, a database management system (such as MySQL) can be used to store the moving target type recognition result and the light signal feature data. Through SQL query statements, the type information and light signal information of the same vehicle are associated. For example, if an SUV is recognized and its vehicle lamp flashing mode is "periodic flashing, with a period of 3 frames", combining the vehicle type and the light signal feature, it is determined that the vehicle is in the "searching for a parking space" state. This identity-state recognition result is stored in the database.

[0028] Obtain the visual image sequence of the target area, and extract the position sequence of the corresponding moving target of the target type-state recognition result in the past preset time length from the visual image sequence of the target area to form the target historical behavior time series dataset; Specifically, based on the target type-state recognition result, the initial position of the target vehicle in the visual image sequence of the target area can be determined. Then, through a frame-by-frame tracking algorithm (such as the Kalman filter), the position change of the vehicle is tracked in consecutive frames. Set the preset time length to 30 seconds. Taking an image sequence of 10 frames per second as an example, the position sequence of the target vehicle within these 300 frames is extracted. For example, if the position coordinates of the vehicle in the first frame are (x1, y1) and the position coordinates in the second frame are (x2, y2), and so on, these position coordinates are arranged in chronological order to form the target historical behavior time series dataset.

[0029] Calculate the instantaneous acceleration of the corresponding moving target according to the target historical behavior time series dataset, and determine the acceleration feature of the moving target based on the instantaneous acceleration; Specifically, the numerical differentiation function gradient in MATLAB software can be used to calculate the time derivative of the vehicle position sequence (the target historical behavior time series dataset). Taking the vehicle position sequence in the x-direction as the input, the instantaneous speed at each moment is calculated, and then the time derivative of the instantaneous speed is calculated to obtain the instantaneous acceleration. For example, if the instantaneous speed of the vehicle at a certain moment is 5 m / s and the instantaneous speed at the next moment is 6 m / s, and the time interval is 0.1 s, then the instantaneous acceleration at this moment is (6 - 5) / 0.1 = 10 m / s². According to the magnitude and sign of the instantaneous acceleration, the acceleration characteristics of the vehicle are judged. If the acceleration is greater than 0, it is determined that the vehicle is accelerating; if the acceleration is less than 0, it is determined that the vehicle is decelerating; if the acceleration is close to 0, it is determined that the vehicle is moving at a constant speed. Among them, the acceleration characteristic of the moving target is any one of the vehicle accelerating, the vehicle decelerating, and the vehicle moving at a constant speed.

[0030] Based on the moving target state characteristic data and the moving target acceleration characteristic, the turning direction of the corresponding moving target is predicted to obtain the moving target direction intention characteristic data; Specifically, a machine learning algorithm (such as Support Vector Machine SVM) can be used for prediction. The vehicle light signal characteristics (such as the headlight flashing pattern, the headlight bright spot distribution) and the acceleration characteristics (such as accelerating, decelerating, moving at a constant speed) are used as input characteristics to construct a training dataset. For example, if the vehicle light signal shows that the left turn signal is flashing and the vehicle is moving at a constant speed, then its turning direction is labeled as "left turn". The SVM model is trained with a large amount of labeled data. During actual prediction, the real-time obtained vehicle light signal and acceleration characteristics are input into the trained model, and the model outputs the vehicle's direction intention characteristic data, such as "going straight", "turning left", "turning right". Among them, the moving target direction intention characteristic data is any one of the vehicle going straight, the vehicle turning left, the vehicle turning right, the vehicle making a U-turn, the vehicle entering a parking space, and the vehicle leaving a parking space.

[0031] Based on the moving target acceleration characteristic and the moving target direction intention characteristic data, the motion intention of the corresponding moving target is fused to obtain the moving target motion intention data.

[0032] Specifically, a series of rules can be preset in a rule engine (such as the Drools rule engine), for example, "if the vehicle is in an accelerating state and the direction intention characteristic is turning left, then it is determined that the moving target motion intention is to quickly turn left and enter the parking space". The acceleration characteristic and the direction intention characteristic of the vehicle are used as the input conditions of the rule, and the rule engine makes inferences according to the preset rules and finally outputs the vehicle's motion intention data. For example, if the vehicle acceleration is positive and the direction intention characteristic is "entering the parking space", then the rule engine determines that the moving target motion intention is "accelerating into the parking space" and outputs this result as the final moving target motion intention data.

[0033] Through multi-dimensional analysis of the vehicle area contour, morphological features, light signal features, and historical behavior time-series data, the present invention can accurately determine the type, state, acceleration characteristics, and driving direction intention of the vehicle. This not only improves the prediction accuracy of vehicle behavior but also effectively copes with the complex and changeable traffic scenarios in the parking lot. This helps to optimize the traffic flow in the parking lot, reduce vehicle conflicts and congestion, improve the operation efficiency and safety of the parking lot, and at the same time provides technical support for the priority passage of special vehicles such as emergency vehicles.

[0034] Preferably, identifying obstacles in the target area according to the moving target motion intention data in step S2 specifically includes: Conduct a feasibility assessment of the moving target motion intention data to obtain a feasibility score for the target moving behavior; Specifically, the Python programming language combined with the machine learning library scikit-learn can be used to evaluate the behavior of the moving target motion intention data. First, extract the key features in the moving target motion intention data, such as vehicle speed, acceleration, and direction intention. Then, use a pre-trained random forest model to score the behavior of each vehicle. For example, set the evaluation indicators of the model to include the smoothness of vehicle driving (acceleration change rate) and the clarity of direction intention (such as whether to change lanes frequently). The model outputs a feasibility score for the target moving behavior, ranging from 0 to 1, where 1 indicates that the behavior is completely reasonable and 0 indicates that the behavior is completely unreasonable. For example, a vehicle driving at a constant speed with a clear direction intention gets a score of 0.9, while a vehicle that frequently brakes suddenly and has a changing direction intention gets a score of 0.3.

[0035] Filter the moving target motion intention data according to the feasibility score of the target moving behavior and a preset behavior rationality threshold to obtain optimized moving target motion intention data; Specifically, the behavior rationality threshold can be set to 0.6, that is, only the moving target motion intention data with a score higher than 0.6 is considered reasonable. Use the Pandas data processing library to filter the moving target motion intention data. For example, from the DataFrame containing all the moving target motion intention data, filter out the rows with a score greater than 0.6 through boolean indexing to obtain the optimized moving target motion intention data.

[0036] Map the optimized moving target motion intention data into a structured semantic description according to the preset semantic mapping rules to obtain motion intention semantic representation data; Specifically, the natural language processing tool NLTK (Natural Language Toolkit) and a predefined semantic mapping rule library can be used to map the optimized mobile target motion intention data into a structured semantic description. For example, the motion intention of a vehicle (such as "accelerate and turn left to enter the parking space") is mapped into a structured semantic description, such as "Vehicle intention: turn left; Action: accelerate; Target: parking space". The semantic mapping rule library defines semantic templates corresponding to various motion intentions. By matching the optimized mobile target motion intention data, motion intention semantic representation data is generated. For example, if the mobile target motion intention is "decelerate and turn right", it is mapped to "Vehicle intention: turn right; Action: decelerate; Target: unknown".

[0037] Obtain the target type - status recognition result, and perform situation association on the motion intention semantic representation data and the target type - status recognition result to obtain the mobile target situation semantic data; Specifically, a database management system (such as MySQL) can be used to store the motion intention semantic representation data and the target type - status recognition result. Through SQL query statements, the intention semantic data and identity - status data of the same vehicle are associated. For example, if the vehicle intention semantic representation is "Vehicle intention: turn left; Action: accelerate; Target: parking space", and the target type - status recognition result is "Vehicle type: SUV; Status: looking for a parking space", then these two pieces of data are combined to generate the mobile target situation semantic data: "Vehicle type: SUV; Status: looking for a parking space; Intention: turn left; Action: accelerate; Target: parking space".

[0038] Perform multi - level semantic layer construction on the mobile target situation semantic data to obtain the scene semantic understanding layer; Specifically, the graph database Neo4j can be used to store and manage the semantic layers. First, create a single - vehicle behavior semantic level, and store the situation semantic data of each vehicle as nodes in the graph database. According to the relative positions and driving directions of the vehicles, construct a fleet behavior semantic level. For example, vehicles that are adjacent and driving in the same direction are associated. According to the areas where the vehicles are located, construct an area behavior semantic level. Further, analyze the overall flow of vehicles in the area and construct a traffic flow behavior semantic level. Finally, identify emergency events (such as a vehicle suddenly braking or making a sharp turn) and construct an emergency event behavior semantic level. Through the construction of these levels, the scene semantic understanding layer is generated.

[0039] Extract the emergency - state features from the scene semantic understanding layer to obtain the mobile target emergency level indicator; Specifically, machine learning algorithms (such as decision trees) can be used to analyze the vehicle behavior in the scene semantic understanding layer. For example, it is set that the emergency state features include the hard braking frequency and sharp turning angle of the vehicle. The decision tree model outputs an emergency level index of the vehicle based on these features, with a range of 1 to 5, where 1 indicates no emergency and 5 indicates a very urgent situation. For example, if the vehicle brakes hard multiple times within a short period and the sharp turning angle is greater than 45 degrees, the model outputs an emergency level index of 5.

[0040] Determine the priority classification of the moving target according to the emergency level index of the moving target and the recognition result of the moving target type; Specifically, a predefined priority rule table can be used to perform priority classification in combination with the vehicle type and the emergency level index. For example, it is set that the moving target with an emergency level index of 5 has the highest priority, and the moving target with an emergency level index of 1 has the lowest priority; at the same time, special vehicles such as fire trucks and ambulances have a higher priority under the same emergency level. By querying the rule table, the emergency level index and type of each vehicle are matched with the priority rules to determine the priority classification of the vehicle. For example, an ambulance with an emergency level index of 5 will be given the highest priority.

[0041] Perform moving obstacle recognition according to the priority classification of the moving target and the scene semantic understanding layer to obtain the regional dynamic obstacle semantic map.

[0042] Specifically, deep learning algorithms (such as the YOLOv5 object detection model) can be used to perform dynamic obstacle recognition in combination with the priority classification information of the moving target. For example, the scene semantic understanding layer of the parking lot is input into the YOLOv5 model, and the model recognizes the position and type of the vehicle, and combines the priority classification information of the moving target to determine which vehicles will become dynamic obstacles. For example, if a high-priority ambulance is driving, other low-priority vehicles are regarded as dynamic obstacles and need to be avoided. Finally, the regional dynamic obstacle semantic map is generated.

[0043] Through the evaluation and optimization of vehicle behavior and the construction of multi-level semantic layers, the present invention can comprehensively understand the traffic scene in the parking lot, including single-vehicle behavior, fleet behavior, regional behavior, traffic flow behavior, and emergency event behavior. This can not only identify dynamic obstacles but also determine their priorities according to the emergency level and type of the vehicle, thus providing a more targeted and flexible basis for subsequent path planning and LED light control. This helps to optimize the passing paths of vehicles in complex dynamic scenarios, especially in emergency situations, reduce conflicts and congestion, and improve the operation efficiency and safety of the parking lot. Preferably, in step S3, the generation of the candidate guiding path set according to the regional dynamic obstacle semantic map is specifically as follows: Obtain the basic topological map of the target area, where the nodes in the basic topological map of the target area represent key positions and the edges represent passable paths; Specifically, AutoCAD software can be used to draw the floor plan of the parking lot, mark the key positions (such as entrances and exits, turning points, parking spaces) in the parking lot as nodes, and mark the passable paths as edges. For example, in a large parking lot, the entrances and exits are marked as node 1 and node 2, and the center point of each parking space is also marked as a node, and the lanes connecting these nodes are marked as edges. The finally generated basic topological map of the target area is saved in the DWG format.

[0044] Calculate the shortest path distance between any two points in the basic topological map of the target area to form an n×n path reachability matrix, where n is the number of key nodes, and the matrix elements represent the shortest distance between nodes or an unreachable mark; Specifically, the basic topological map of the target area can be imported into networkx in Python and represented as a graph structure. Use the all_pairs_dijkstra_path_length function provided by networkx to calculate the shortest path lengths between all node pairs in the graph. For example, for a parking lot topological map with 100 key nodes, this function will generate a 100×100 path reachability matrix. Each element in the matrix represents the shortest distance between the corresponding node pair. If two nodes are unreachable, it is marked as infinity (inf). Through the above steps, the shortest path information between any two points in the parking lot can be obtained quickly.

[0045] Perform 3D spatial modeling on the path reachability matrix and the basic topological map of the target area to obtain the digital twin model of the target area; Specifically, the basic topological map of the target area can be imported into Unity, and the 2D topological map can be converted into a 3D model using Unity's 3D modeling tools. For example, represent the lane as a 3D road model with a certain width and height, and represent the node as a specific position point on the road. Use the shortest path information in the path reachability matrix as the input data for the navigation mesh (NavMesh) to generate the 3D navigation mesh of the parking lot. This navigation mesh not only contains the spatial structure information of the parking lot but also the path reachability information, providing a 3D virtual environment for subsequent path planning and vehicle navigation, that is, the digital twin model of the target area.

[0046] Perform obstacle rasterization on the regional dynamic obstacle semantic map to obtain the target area environmental occupancy raster map; Specifically, the regional dynamic obstacle semantic map can be imported into MATLAB. This map contains the position and type information of dynamic obstacles such as vehicles and pedestrians. The imbinarize function in the image processing toolbox of MATLAB is used to convert the semantic map into a binary image, where the obstacle area is 1 and the non-obstacle area is 0. The imresize function is used to adjust the binary image to the same resolution as the target area environmental occupancy grid map. For example, the parking lot is divided into grids of 10 cm × 10 cm to generate a rasterized target area environmental occupancy grid map. In this grid map, the value of each grid represents whether there is an obstacle at that position.

[0047] Obtain the current positions and target positions of each moving target in the target area; convert the current positions and target positions of each moving target in the target area into a set of position points in the grid coordinate system of the target area environmental occupancy grid map as the starting and ending points of path planning, thereby obtaining the target position coordinate set; Specifically, the current positions and target positions of each vehicle in the parking lot can be obtained through a vehicle positioning system (such as GPS) and a parking lot management system (such as a parking space reservation system). For example, the GPS device installed in the vehicle can send its current position coordinates (longitude and latitude) to the parking lot management system in real time. At the same time, the parking lot management system records the position of the reserved parking space of the vehicle as the target position. Converting these position information into the grid coordinate system of the target area environmental occupancy grid map can be completed through a coordinate conversion formula. For example, assuming that the lower left corner of the parking lot is the origin (0, 0) of the grid coordinate system, according to the GPS coordinates of the vehicle and the geographical coordinate range of the parking lot, the set of position points of the vehicle in the grid coordinate system is calculated. These sets of position points are used as the starting and ending points of path planning to form the target position coordinate set.

[0048] Based on the target area digital twin model and the target area environmental occupancy grid map, calculate the passing cost of each passable grid in the target area environmental occupancy grid map, and generate a path planning basic cost map according to the passing cost. Among them, the specific calculation formula for calculating the passing cost of each passable grid is as follows: ; where L is the passing cost, K is the distance between the grid and the obstacle, Q is the curvature of the path where the grid is located, Y is the passing priority of the grid, is the distance weight coefficient, is the curvature weight coefficient, is the priority weight coefficient; Specifically, the calculation formula of the passing cost can be defined , where, is the distance weight coefficient, which is used to adjust the influence degree of the distance between the grid and the obstacle on the passing cost, is the curvature weight coefficient, which is used to adjust the influence degree of the curvature of the path where the grid is located on the passing cost. is the priority weight coefficient, which is used to adjust the influence degree of the passing priority of the grid on the passing cost. For example, set = 0.5, = 0.3, = 0.2. For each passable grid, calculate the distance K from it to the nearest obstacle, obtain the curvature Q of the path where the grid is located through the path curvature information in the digital twin model, and determine the Y value according to the passing priority of the grid (for example, the main road has a higher priority). Then, substitute these values into the formula to calculate the passing cost L of each grid, and store the passing costs of all grids in a two-dimensional array to generate the basic cost map for path planning. This map reflects the passing difficulty of each grid in the parking lot.

[0049] Perform path search based on the basic cost map for path planning and the target position coordinate set to obtain a set of candidate guiding paths.

[0050] Specifically, the basic cost map for path planning can be imported into networkx and represented as a weighted graph, with the passing cost of each grid as the weight of the edge. Then, for each starting point and ending point in the target position coordinate set, use the dijkstra_path function provided by networkx for path search. This function will find the shortest path from the starting point to the ending point according to the weights in the cost map. For example, for a vehicle from the entrance (starting point) to the reserved parking space (ending point), the Dijkstra algorithm will search for a path that avoids obstacles and has the minimum passing cost, and use this path as one of the candidate guiding paths. Through the above steps, a set of candidate guiding paths is generated for each vehicle in the parking lot.

[0051] By combining the basic topological map of the target area with the semantic map of dynamic obstacles and introducing the passing cost calculation formula, the present invention can comprehensively consider factors such as the distance between the grid and the obstacle, the path curvature, and the passing priority, and dynamically evaluate the passing cost of each passable grid. This can not only generate multiple candidate guiding paths, but also ensure the feasibility and safety of the paths, while taking into account the optimization of the paths and the avoidance of dynamic obstacles. This enables the system to quickly respond to the driving needs of vehicles in a complex and changeable parking lot environment, and effectively avoid congestion and collision risks.

[0052] Preferably, in step S3, generating the initial guiding path map based on the set of candidate guiding paths is specifically as follows: Obtain the environmental occupancy grid map of the target area, and calculate the grid occupancy rate of the target area based on the environmental occupancy grid map of the target area; Specifically, a two-dimensional map of the parking lot can be collected, imported into MATLAB, and divided into 10cm×10cm grids. Use MATLAB's imread function to read the regional dynamic obstacle semantic map (binary image), where the obstacle area is 1 and the non-obstacle area is 0. Use the im2bw function to convert the image into a binary image to ensure that there are only two states in the image: obstacles and non-obstacles. Count the proportion of obstacles in each grid and calculate the grid occupancy rate of the target area. For example, if the obstacle pixel ratio in a grid is 30%, the occupancy rate of the grid is 0.3. Finally, the grid occupancy distribution data is generated.

[0053] Obtain historical traffic data of the target area, and determine the distribution data of mobile targets in the current target area based on the environmental perception data of the target area; Specifically, an SQL database can be used to store historical traffic data of a parking lot. This data includes information such as vehicle entry and exit records and vehicle dwelling time in the parking lot during different time periods every day. Through SQL query statements, historical traffic data within a specific time period can be extracted. For example, query the vehicle traffic data from 8 am to 10 am every day in the past month. At the same time, based on the environmental perception data of the target area (such as the current location of the vehicle and the target location), the data processing tools in MATLAB or Python (such as Pandas) are used to determine the distribution data of mobile targets in the current target area. For example, count the number of vehicles in each area of ​​the current parking lot and combine these data with historical traffic data.

[0054] The congestion index of each candidate guiding path in the candidate guiding path set is calculated according to the target area grid occupancy rate, the target area historical traffic data and the current target area mobile target distribution data, and a path congestion index table is generated, wherein the specific calculation method of calculating the congestion index of each candidate guiding path in the candidate guiding path set is as follows: ; Among them, W is the congestion index, S is the grid occupancy rate, P is the weighted average of historical traffic, M is the current mobile target density, is the grid occupancy weight coefficient, is the historical traffic weight coefficient, is the current moving target density weight coefficient; Specifically, based on the grid occupancy rate distribution data, the proportion S of occupied grids on each candidate guiding path can be calculated. For example, if a path passes through 100 grids and 30 of them are occupied, then S = 0.3. Then, extract the historical traffic data of this path from the SQL database and calculate the historical traffic weighted average P. For example, set the traffic weights of this path in the past month to be 0.1, 0.2, 0.3, etc., and calculate the weighted average. At the same time, according to the current vehicle distribution data, calculate the vehicle density M on the current path. For example, if there are 10 vehicles on the current path and the path length is 100 meters, then M = 0.1 (vehicles / meter). Set the weight coefficients = 0.4, = 0.3, = 0.3, substitute into the formula , and calculate the congestion index of each path. Finally, generate a path congestion index table. Among them, S is the grid occupancy rate, representing the proportion of occupied grids on the path, P is the historical traffic weighted average, reflecting the traffic situation of the path based on historical data, M is the current moving target density, representing the vehicle density on the current path, is the grid occupancy rate weight coefficient, used to adjust the influence degree of the grid occupancy rate on the congestion index, is the historical traffic weight coefficient, used to adjust the influence degree of the historical traffic on the congestion index, is the current moving target density weight coefficient, used to adjust the influence degree of the current moving target density on the congestion index.

[0055] Evaluate the path safety interval of each candidate guiding path in the candidate guiding path set according to the preset path safety interval scoring criterion, and obtain a path safety score table. Among them, the factors considered in the path safety interval include the minimum distance between the path and obstacles, the number of path intersections, and the path visibility range; Specifically, a safety interval scoring criterion can be defined, considering three factors: the minimum distance between the path and obstacles, the number of path intersections, and the path visibility range. For example, set the weight of the minimum distance between the path and obstacles to be 0.4, the weight of the number of intersections to be 0.3, and the weight of the visibility range to be 0.3. For each path, calculate its minimum distance from the obstacle (unit: meter), count the number of intersections on the path, and evaluate the visibility range of the path (unit: meter). For example, the minimum distance between a path and an obstacle is 2 meters, the number of intersections is 3, and the visibility range is 10 meters. Calculate the path safety score according to the weights and generate a path safety score table.

[0056] Comprehensively sort the candidate guiding paths in the candidate guiding path set according to the path safety score table and the path congestion index table to obtain a comprehensive ranking table of guiding paths; Specifically, the path safety score and the congestion index can be used as two main indicators, with weights set at 0.6 and 0.4 respectively. For example, for a path, its safety score is 80 points (out of 100), and the congestion index is 0.2 (the lower the better). Calculate the comprehensive score: Comprehensive score = 0.6 × Safety score - 0.4 × Congestion index. Through the above steps, score all candidate paths, and sort them according to the comprehensive score to generate a comprehensive ranking table of the guiding paths.

[0057] Select the guiding path with the best comprehensive ranking for each target point in the set of target position coordinates according to the comprehensive ranking table of the guiding paths, and form an initial guiding path map including a sequence of path points, turning point markers, and guiding instructions.

[0058] Specifically, the optimal path corresponding to each target point can be extracted from the comprehensive ranking table of the guiding paths. For example, for a target point, its optimal path is the path with the highest ranking. Then, extract the sequence of path points, turning point markers, and guiding instruction information of these optimal paths. Using the Matplotlib plotting library, take the occupancy grid map of the target area environment as the background, and plot the optimal paths on the map to form an initial guiding path map including a sequence of path points, turning point markers, and guiding instructions. This map intuitively shows the optimal driving path of the vehicle from the current position to the target position.

[0059] By calculating the grid occupancy rate, combining the historical traffic flow data and the current vehicle distribution data of the target area, the present invention can accurately evaluate the congestion situation of each candidate path, thereby effectively avoiding congested areas and improving the vehicle passing efficiency. At the same time, based on the path safety interval scoring criterion, it can fully consider the safety of the path, including factors such as the distance from obstacles, the number of intersections, and the visibility range, to ensure the safety of vehicle driving. This enables the system to select the optimal guiding path for each target point in a complex and changeable parking lot environment. The generated initial guiding path map can not only effectively guide the vehicle to quickly reach the destination, but also significantly reduce the collision risk and improve the overall operation efficiency and safety of the parking lot.

[0060] Preferably, in step S3, detecting the conflicting paths in the initial guiding path map and performing path avoidance processing on the conflicting paths is specifically as follows: Map the priorities of the moving targets in the regional dynamic obstacle semantic map to the corresponding paths in the initial guiding path map to obtain a priority path mapping table; Specifically, the priority information of each vehicle can be extracted from the regional dynamic obstacle semantic map, and this information includes vehicle types (such as ambulances, fire trucks) and emergency level indicators. For example, the priority of an ambulance is marked as the highest (priority level 1), while the priority of an ordinary vehicle is 5. Then, the initial guiding path map is imported into a Pandas DataFrame, and each path has a unique path ID. By the current location and target location of the vehicle, the priority information of the vehicle is associated with the path ID to generate a priority path mapping table. For example, there is an ambulance with a priority of 1 on the path with path ID 101, and an ordinary vehicle with a priority of 5 on the path with path ID 102. Finally, a priority path mapping table is generated.

[0061] Construct a path scheduling strategy for the target area according to the priority path mapping table; Specifically, a path scheduling strategy for the target area can be constructed using the graph database Neo4j according to the priority path mapping table. First, the priority path mapping table is imported into Neo4j, and each path and its corresponding moving target priority are stored as nodes in the graph database. Utilizing the graph algorithm function of Neo4j, the paths are sorted and scheduled according to the moving target priority. For example, the rule is set that the paths of high-priority vehicles pass first, and the paths of low-priority vehicles need to be adjusted. Through the Cypher query language, a path scheduling strategy can be defined, such as "if there is a high-priority vehicle on a path, the path of the low-priority vehicle needs to be re-planned". Through the above steps, Neo4j can dynamically adjust the path scheduling strategy according to the moving target priority.

[0062] Identify the path conflict points in the path scheduling strategy for the target area that cause collisions or blockages of moving targets, and generate conflict identification data including the conflict location, time, and the path IDs involved; Specifically, the path information in the path scheduling strategy for the target area can be imported into MATLAB, including the starting point, ending point, sequence of path points, and moving target priority of the path. The conflict points are detected by calculating the spatial and temporal overlaps between paths. For example, if two paths pass through the same grid point within the same time period, it is considered that there is a conflict. MATLAB can utilize its matrix operation function to quickly calculate the intersection points and overlapping areas between paths. For each conflict point, conflict identification data including the conflict location (grid coordinates), conflict time (timestamp), and the path IDs involved is generated. For example, the conflict identification data shows that path ID 101 and path ID 102 conflict at location (5, 10) at time 14:30.

[0063] Apply a preset priority-based conflict resolution strategy to each conflict according to the conflict identification data to generate a path conflict resolution plan. The preset priority-based conflict resolution strategy includes path replanning, timing adjustment, or yielding control. The time limit for path replanning is no more than 5 seconds each time. The maximum adjustment time range for timing adjustment is from -30 seconds to +30 seconds. The time range for yielding control is 5 seconds - 60 seconds. Specifically, the conflict identification data can be imported into Python. According to the preset conflict resolution strategies (such as path replanning, timing adjustment, or yielding control) and the priorities of moving targets, a decision tree algorithm is used to generate a path conflict resolution plan. For example, if a high-priority vehicle and a low-priority vehicle conflict at the same location, the decision tree will recommend replanning the path of the low-priority vehicle or adjusting the passing time of the low-priority vehicle. The specific strategy can be implemented through a rule engine (such as Drools). For example, "if the conflict involves a high-priority vehicle, the low-priority vehicle yields". Through the above steps, a specific solution is generated for each conflict and these solutions are stored in the path conflict resolution plan table.

[0064] Generate an optimal guiding path map according to the path conflict resolution plan and the path scheduling strategy for the target area.

[0065] Specifically, the path conflict resolution plan and the path scheduling strategy can be combined to adjust each path. For example, according to the conflict resolution plan, replan the path of the low-priority vehicle or adjust the passing time. Import the adjusted path information into Matplotlib. Using the occupancy grid map of the target area environment as the background, plot the path point sequence, turning point markers, and guiding indications of each path. For example, for an adjusted path, different colored lines can be used to represent the path, arrows can be used to mark the turning points, and text can be used to label the guiding indications (such as "turn left", "go straight"). Finally, the generated optimal guiding path map will clearly show the optimal driving paths of each vehicle.

[0066] By mapping the priorities of moving targets to paths and constructing a path scheduling strategy, the present invention can identify conflict points that are likely to cause collisions or blockages and generate detailed conflict identification data. Based on this data, preset conflict resolution strategies such as path replanning, timing adjustment, or yielding control are applied to dynamically adjust the paths, thereby generating an optimal guiding path map. This can significantly reduce the traffic conflict risk in the parking lot, improve the vehicle passing efficiency, and ensure traffic smoothness.

[0067] Preferably, step S4 includes the following steps: Step S41: Obtain the floor plan of the target area, divide the floor plan of the target area into grids according to a fixed size, and correspond each path point in the optimal guiding path map to the grid cells in the divided floor plan of the target area, generating a path-grid correspondence table containing grid IDs and path IDs; Specifically, the floor plan of the target area can be opened in AutoCAD, and its drawing tools can be used to divide the floor plan into grids according to a fixed size (such as 1 meter × 1 meter). Each grid cell is assigned a unique grid ID. The path point data in the optimal guiding path map is exported as a CSV file, which contains the coordinates and path IDs of the path points. Use a Python script to read the CSV file and correspond each path point to the corresponding grid cell through coordinate matching. For example, a path point with coordinates (2.5, 3.5) will be mapped to the grid cell with grid ID (3, 4). Finally, a path-grid correspondence table is generated, which contains the mapping relationship between grid IDs and path IDs.

[0068] Step S42: Calculate the path similarity of adjacent grid cells based on the path-grid correspondence table, perform clustering on the grid cells to obtain clustering results, and divide blocks with similar guiding functions according to the clustering results, generating LED light control function blocks; Specifically, the path ID information of each grid cell can be extracted from the path-grid correspondence table. Then, calculate the path similarity between adjacent grid cells, for example, by calculating the overlap degree of path IDs. If two adjacent grid cells share multiple path IDs, their similarity is relatively high. Use the K-Means clustering algorithm in scikit-learn to divide the grid cells into multiple clusters according to the path similarity. For example, set the number of clusters to 5, and the algorithm will divide the grid cells into 5 blocks with similar guiding functions according to the similarity. Each block will be assigned a function label, such as "guiding function" or "warning function". Finally, LED light control function blocks are generated.

[0069] Step S43: Obtain the actual LED light position distribution map, map the LED light control function blocks to the actual LED light positions based on the actual LED light position distribution map, determine the control zone to which each LED light belongs, and assign control priorities according to the functional characteristics of the control zone, generating a regional LED control zone map containing LED light IDs, zone IDs, and control priorities, where the functional characteristics of the control zone are at least any one or more of guiding function, warning function, no-entry function, priority passage function, and emergency evacuation function; Specifically, the floor plan of the target area and the distribution map of LED lamp positions can be imported into ArcGIS. The distribution map of LED lamp positions contains the coordinates and IDs of each LED lamp. Then, the LED lamp control function blocks are imported into ArcGIS, and each LED lamp is matched with the nearest function block through a spatial analysis tool. For example, if an LED lamp is located within a "guidance function" block, the LED lamp is assigned to that block. According to the characteristics of the function blocks (such as guidance, warning, no entry), control priorities are assigned to each LED lamp. For example, the LED lamps of the "emergency evacuation function" have the highest priority. Finally, a regional LED control zoning map is generated, which contains LED lamp IDs, zoning IDs, and control priority information.

[0070] Step S44: Generate a synchronous LED control instruction set according to the regional LED control zoning map; configure the dynamic effect parameters of the LED lamps according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table.

[0071] Specifically, the control priorities and functional characteristics of each LED lamp can be extracted from the regional LED control zoning map. Then, a synchronous LED control instruction set is generated according to the functional characteristics. For example, for the LED lamps of the "guidance function", the instruction set includes "blinking frequency is 2 Hz, color is green"; for the LED lamps of the "warning function", the instruction set includes "blinking frequency is 4 Hz, color is yellow". Use MATLAB to write a control algorithm to dynamically adjust the parameters of the LED lamps, such as brightness, color, and blinking frequency, according to the control instruction set. Finally, an LED dynamic display parameter table is generated, which contains the dynamic effect parameters of each LED lamp.

[0072] By meshing the floor plan of the target area and mapping it with the optimal guiding path map, the present invention can accurately locate the grid cells corresponding to each path point, and then divide the blocks with similar guiding functions through cluster analysis, providing a scientific basis for the zoning control of LED lamps. By combining the actual LED lamp position distribution map, the LED lamps can be accurately assigned to each function block, and control priorities are assigned according to the functional characteristics to generate a regional LED control zoning map. This not only improves the control flexibility and response speed of the LED lamps, but also can dynamically adjust the lighting effects according to different functional requirements (such as guidance, warning, no entry, priority passage, emergency evacuation), significantly enhancing the intuitiveness and accuracy of vehicle guidance in the parking lot.

[0073] Preferably, the present invention provides an intelligent guiding control system for LED lamps, which is used to execute the intelligent guiding control method for LED lamps as described above. The intelligent guiding control system for LED lamps includes: An environment perception module, which is used to obtain the environment perception data of the target area; An intention and obstacle recognition module, which is used to recognize the motion intentions of each moving target in the target area based on the environmental perception data of the target area, so as to obtain moving target motion intention data; and recognize the obstacles in the target area according to the moving target motion intention data to generate a regional dynamic obstacle semantic map; A guiding path planning module, which is used to generate a candidate guiding path set according to the regional dynamic obstacle semantic map; generate an initial guiding path map based on the candidate guiding path set; detect the conflicting paths in the initial guiding path map, and perform path avoidance processing on the conflicting paths to obtain an optimal guiding path map; An LED parameter configuration module, which is used to perform instruction partition collaborative control on the LED lamp system according to the optimal guiding path map to obtain a synchronous LED control instruction set; configure the dynamic effect parameters of the LED lamp according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table; An LED guiding control module, which is used to perform a lighting effect rendering guiding control operation on the LED lamp system according to the LED dynamic display parameter table.

[0074] Preferably, the present invention also provides an LED lamp for executing the above-mentioned LED lamp intelligent guiding control method.

[0075] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to cover all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0076] The above are only specific embodiments of the present invention, which enable those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for intelligent guidance control of LED lights, characterized in that: The following steps are involved: Step S1: Acquire target area environmental perception data; Step S2: identifying the motion intention of each mobile target in the target area based on the target area environment perception data, and obtaining the motion intention data of the mobile target; identifying obstacles in the target area according to the motion intention data of the mobile target, and generating a regional dynamic obstacle semantic map; Step S3: generating a candidate guidance path set according to the regional dynamic obstacle semantic map; generating an initial guidance path map based on the candidate guidance path set; detecting conflicting paths of the initial guidance path map, and performing path avoidance processing on the conflicting paths to obtain an optimal guidance path map; Step S4: performing instruction partitioning collaborative control on the LED light system according to the optimal guidance path diagram to obtain a synchronous LED control instruction set; configuring the dynamic effect parameters of the LED light according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table; Step S5: performing lighting effect rendering guidance control operations on the LED lamp system according to the LED dynamic display parameter table.

2. The LED lamp intelligent guidance control method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: collecting moving target heat map data through the infrared sensor array, and collecting target area obstacle distance data through the ultrasonic sensor array; Step S12: using a microwave radar to collect the moving target's speed, and using a camera to collect visual image data of the target area to form a visual image sequence of the target area, wherein the microwave radar sampling rate ranges from 10 Hz to 100 Hz, and the camera frame rate ranges from 15 fps to 60 fps; Step S13: performing time stamp calibration on the moving target heat map data and the target area obstacle distance data to obtain moving target-obstacle time alignment data; Step S14: construct a three-dimensional space coordinate system; based on the three-dimensional space coordinate system, integrate the moving target speed, the moving target-obstacle time alignment data and the target area visual image sequence to obtain the target area environment perception data.

3. The LED lamp intelligent guidance control method according to claim 1, characterized in that: The step S2 of identifying the motion intention of each moving target in the target area based on the target area environment perception data is specifically as follows: Extracting the area contours of each moving target in the target area from the target area environment perception data to obtain a moving target candidate area map; Extracting the morphological features of the mobile target in the mobile target candidate area map to obtain a mobile target type feature vector, and matching and classifying the mobile target type feature vector with a preset mobile target type feature vector library to obtain a mobile target type recognition result; Identify the motion state of the mobile target based on the target area environment perception data to form mobile target state feature data; Determine a target type-state recognition result according to the mobile target type recognition result and the mobile target state characteristic data; Acquire a target area visual image sequence, extract a position sequence of a mobile target corresponding to a target type-state recognition result within a preset time length in the past from the target area visual image sequence, and form a target historical behavior time series data set; Calculating the instantaneous acceleration of the corresponding moving target according to the target historical behavior time series data set, and determining the acceleration characteristics of the moving target according to the instantaneous acceleration; Predicting the turning direction of the corresponding moving target according to the moving target state feature data and the moving target acceleration feature, and obtaining the moving target direction intention feature data; The motion intention of the corresponding moving target is fused according to the moving target acceleration characteristics and the moving target direction intention feature data to obtain the moving target motion intention data.

4. The LED lamp intelligent guidance control method according to claim 1, characterized in that: The step S2 of identifying obstacles in the target area according to the moving target motion intention data is specifically as follows: Conducting a feasibility assessment on the movement intention data of the moving target to obtain a feasibility score for the target's movement behavior; The moving target movement intention data is screened according to the feasibility score of the moving behavior of the target and the preset behavior rationality threshold to obtain the optimized moving target movement intention data; According to a preset semantic mapping rule, the optimized moving target motion intention data is mapped into a structured semantic description to obtain motion intention semantic representation data; Obtaining target type-state recognition results, contextually associating motion intention semantic representation data with the target type-state recognition results, and obtaining moving target context semantic data; Construct a multi-level semantic layer for the mobile target context semantic data to obtain a scene semantic understanding layer; Extract emergency state features from the scene semantic understanding layer to obtain the emergency level index of the moving target; Determine the priority classification of mobile targets according to the mobile target urgency index and the mobile target type identification result; According to the mobile target priority classification and scene semantic understanding layer, mobile obstacles are identified to obtain the regional dynamic obstacle semantic map.

5. The LED lamp intelligent guidance control method according to claim 1, characterized in that: The step S3 generates a candidate guide path set according to the regional dynamic obstacle semantic map, specifically: Obtaining a basic topological map of the target area, wherein nodes in the basic topological map of the target area represent key locations and edges represent traversable paths; Calculate the shortest path distance between any two points in the target area basic topology map to form an n×n path accessibility matrix, where n is the number of key nodes and the matrix elements represent the shortest distance between nodes or unreachable marks; Conduct three-dimensional spatial modeling of the path accessibility matrix and the basic topological map of the target area to obtain a digital twin model of the target area; Perform obstacle rasterization on the regional dynamic obstacle semantic map to obtain the target area environment occupancy raster map; Obtain the current position and target position of each mobile target in the target area; convert the current position and target position of each mobile target in the target area into a set of position points in a grid coordinate system in the target area environment occupancy grid map as the starting point and end point of the path planning, thereby obtaining a target position coordinate set; Based on the target area digital twin model and the target area environmental occupancy grid map, the pass cost of each passable grid in the target area environmental occupancy grid map is calculated, and the path planning basic cost map is generated according to the pass cost, wherein the specific calculation formula for calculating the pass cost of each passable grid is as follows: ； Among them, L is the cost of passage, K is the distance between the grid and the obstacle, Q is the curvature of the path where the grid is located, and Y is the passage priority of the grid. is the distance weight coefficient, is the curvature weight coefficient, is the priority weight coefficient; A path search is performed based on the path planning basic cost map and the target position coordinate set to obtain a set of candidate guided paths.

6. The LED lamp intelligent guidance control method according to claim 1, characterized in that: The step S3 generates an initial guidance path graph based on the candidate guidance path set, specifically: Obtaining a target area environment occupancy grid map, and calculating the target area grid occupancy rate based on the target area environment occupancy grid map; Obtain historical traffic data of the target area, and determine the distribution data of mobile targets in the current target area based on the environmental perception data of the target area; The congestion index of each candidate guiding path in the candidate guiding path set is calculated according to the target area grid occupancy rate, the target area historical traffic data and the current target area mobile target distribution data, and a path congestion index table is generated, wherein the specific calculation method of calculating the congestion index of each candidate guiding path in the candidate guiding path set is as follows: ； Among them, W is the congestion index, S is the grid occupancy rate, P is the weighted average of historical traffic, M is the current mobile target density, is the grid occupancy weight coefficient, is the historical traffic weight coefficient, is the current moving target density weight coefficient; Evaluate the path safety interval of each candidate guiding path in the candidate guiding path set according to a preset path safety interval scoring criterion to obtain a path safety scoring table, wherein the factors considered in the path safety interval include the minimum distance between the path and the obstacle, the number of path intersections, and the path visibility range; Comprehensively sorting the candidate guiding paths in the candidate guiding path set according to the path safety score table and the path congestion index table to obtain a comprehensive ranking table of guiding paths; According to the comprehensive ranking table of guidance paths, the guidance path with the best comprehensive ranking is selected for each target point in the target position coordinate set, forming an initial guidance path diagram including a path point sequence, turning point marks and guidance instructions.

7. The LED lamp intelligent guidance control method according to claim 1, characterized in that: The step S3 detects the conflicting paths of the initial guidance path graph and performs path avoidance processing on the conflicting paths, specifically: Map the priorities of mobile targets in the regional dynamic obstacle semantic graph to the corresponding paths in the initial guidance path graph to obtain a priority path mapping table; Construct a target area path scheduling strategy based on the priority path mapping table; Identify the path conflict points in the target area path scheduling strategy that cause collision or blockage of mobile targets, and generate conflict identification data including conflict location, time and involved path ID; Applying a preset priority-based conflict resolution strategy to each conflict according to the conflict identification data to generate a path conflict resolution solution, wherein the preset priority-based conflict resolution strategy includes path replanning, timing adjustment or yield control, the path replanning time limit is no more than 5 seconds each time, the maximum adjustment time range of the timing adjustment is -30 seconds to +30 seconds, and the time range of the yield control is 5 seconds to 60 seconds; Generate an optimal guidance path map based on the path conflict resolution method and the target area path scheduling strategy.

8. The LED lamp intelligent guidance control method according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: obtaining a target area plan, dividing the target area plan into grids of a fixed size, making one-to-one correspondence between each path point in the optimal guidance path map and a grid unit in the divided target area plan, and generating a path grid correspondence table including grid IDs and path IDs; Step S42: Calculate the path similarity of adjacent grid units based on the path grid correspondence table, cluster the grid units to obtain clustering results, divide blocks with similar guidance functions according to the clustering results, and generate LED light control function blocks; Step S43: obtaining an actual LED lamp position distribution map, mapping the LED lamp control function block with the actual LED lamp position based on the actual LED lamp position distribution map, determining the control partition to which each LED lamp belongs, assigning a control priority according to the functional characteristics of the control partition, and generating a regional LED control partition map including an LED lamp ID, a partition ID and a control priority, wherein the functional characteristics of the control partition are at least any one or more of a guiding function, a warning function, a prohibition function, a priority passage function, and an emergency evacuation function; Step S44: Generate a synchronous LED control instruction set according to the regional LED control partition diagram; configure the dynamic effect parameters of the LED lamp according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table.

9. An intelligent guidance control system for LED lights, characterized in that: Used to execute the LED lamp intelligent guidance control method according to claim 1, the LED lamp intelligent guidance control system comprises: Environmental perception module, used to obtain environmental perception data of the target area; The intention and obstacle recognition module is used to identify the movement intention of each mobile target in the target area based on the target area environmental perception data, and obtain the movement intention data of the mobile target; identify the obstacles in the target area according to the movement intention data of the mobile target, and generate a regional dynamic obstacle semantic map; The guidance path planning module is used to generate a candidate guidance path set according to the regional dynamic obstacle semantic map; generate an initial guidance path map based on the candidate guidance path set; detect conflicting paths in the initial guidance path map, and perform path avoidance processing on the conflicting paths to obtain an optimal guidance path map; The LED parameter configuration module is used to perform command partitioning and coordinated control of the LED light system according to the optimal guidance path diagram to obtain a synchronous LED control instruction set; configure the dynamic effect parameters of the LED light according to the synchronous LED control instruction set to obtain an LED dynamic display parameter table; The LED guidance control module is used to perform lighting effect rendering guidance control operations on the LED light system according to the LED dynamic display parameter table.

10. An LED lamp, characterized in that: Used to execute the LED lamp intelligent guidance control method as described in claim 1.

Images