Unmanned Aerial Vehicle (UAV) Spraying System for Snow Cover Recognition and Dynamic Path Planning at Interchanges

Abstract

This invention discloses a drone-based snow removal system for grade-separated roads, focusing on snow accumulation identification and dynamic path planning. It belongs to the fields of intelligent transportation and municipal sanitation technology. The system includes modules for snow accumulation identification, CAD drawing parsing, take-off and landing point management, material and power estimation, dynamic task splitting, path generation and sorting, resupply determination, and visualization. A lightweight FLSD model is used to identify snow accumulation in real time at the edge, and combined with a dynamic consumption model and a comprehensive cost function, dynamic path planning for spraying is achieved based on snow distribution, resource status, and safety margin. The system supports flexible task splitting, intelligent return-to-home and resupply decisions, and provides full-process visualization output. Compared to traditional methods, this invention significantly improves the accuracy, safety, and resource utilization efficiency of snow removal on complex grade-separated roads, and is suitable for intelligent winter maintenance in high-risk areas such as elevated roads and ramps.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation and municipal sanitation technology, specifically relating to a drone spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads. It integrates computer vision, edge computing, path planning and autonomous decision-making technologies, and is suitable for snow accumulation recognition and precise spraying operations in complex grade-separated road scenarios. Background Technology

[0002] Currently, winter snow removal operations on roads mainly rely on manual sweeping, salt trucks, or large mechanical snow removal equipment. These traditional methods have the advantages of high efficiency and wide coverage on highways or open sections, but they reveal significant limitations in complex scenarios such as overpasses, multi-ramp intersections, and elevated curves: on the one hand, large equipment has difficulty entering narrow areas or areas with large elevation differences, resulting in blind spots and safety risks; on the other hand, existing operation modes generally adopt fixed routes or timed scheduling strategies, lacking the ability to perceive the actual snow distribution in real time, and failing to achieve "on-demand spraying," leading to problems such as waste of de-icing agents, untimely snow removal, or excessive operations.

[0003] In recent years, some studies have attempted to introduce visual technologies to identify road snow accumulation, such as image analysis methods based on vehicle-mounted or fixed cameras. However, these solutions are mostly offline processing or only used for post-event evaluation, failing to deeply integrate with mobile operation platforms such as drones, and even more so failing to form a closed-loop control system of "perception-decision-execution". In addition, existing drone spraying systems typically use preset flight routes, with static path planning that cannot dynamically adjust the task sequence and flight strategy based on real-time snow accumulation, remaining battery power, or spraying material. This results in low resource utilization, high risk of mission interruption, and difficulty in adapting to the actual needs of complex topologies and variable environments in grade-separated roads.

[0004] In summary, existing technologies urgently need to address the following core issues:

[0005] 1. How to achieve real-time, high-precision identification of road snow accumulation under complex lighting and background interference on a drone platform with limited computing power; 2. How to deeply couple snow accumulation perception results with task planning to build a dynamic and reconfigurable spraying path generation mechanism; 3. How to comprehensively consider power consumption, material consumption, return safety and operation efficiency during flight to achieve intelligent task scheduling and flexible execution under resource constraints.

[0006] Therefore, there is an urgent need for a drone spraying system that integrates multimodal perception, edge intelligence, dynamic path planning, and autonomous decision-making to improve the intelligence, precision, and safety of winter snow removal operations in complex grade-separated road scenarios. Summary of the Invention

[0007] To address the aforementioned issues, this invention provides a drone-based snow removal system for grade-separated roads, enabling snow accumulation identification and dynamic path planning. The system aims to construct an intelligent operational framework integrating real-time snow perception, dynamic task planning, adaptive resource scheduling, and closed-loop execution feedback. By fusing infrared and visible light multimodal visual perception modules, the system achieves high-precision, low-latency identification of snow-covered areas on complex grade-separated roads at the drone's edge. It automatically generates structured flight segments based on CAD drawings and dynamically generates and optimizes spraying paths based on multi-dimensional constraints such as snow distribution, remaining battery power, spraying materials, and return-to-home safety margin. This supports flexible task splitting and intelligent resupply decisions. Ultimately, it achieves the operational goals of "on-demand response, precise snow removal, and optimal resource utilization," significantly improving the efficiency, safety, and intelligence of snow removal operations in complex road network scenarios such as elevated roads, ramps, and curves.

[0008] The technical solution adopted in this invention is as follows:

[0009] A drone spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads includes: a road snow accumulation recognition module, a path generation and sorting module, and a material and power estimation module.

[0010] The road snow recognition module is based on a lightweight FLSD model, which enables low-latency and high-precision real-time snow detection under complex lighting conditions at the edge of the drone, and outputs the real-time snow detection results to the path generation and sorting module.

[0011] The path generation and sorting module uses a comprehensive cost function that integrates entry distance, return distance and resource safety margin, based on the current starting position and the set of flight segments, and dynamically generates continuously executed spraying paths through an improved greedy algorithm.

[0012] The materials and power estimation module, based on flight segment geometry parameters and environmental factors, uses a dynamically adjusted consumption model to accurately predict the demand for spraying materials and battery power in real time, providing a resource constraint basis for mission scheduling and resupply decisions.

[0013] Furthermore, the road snow recognition module is based on the lightweight model FLSD built on the improved YOLOv10. It introduces a depthwise separable convolution + CSP structure in the backbone network, introduces an attention mechanism of multi-scale feature fusion of spatial attention and channel attention in the feature extraction stage, and adopts structured pruning, INT8 weight quantization and operator fusion for real-time edge inference in the edge inference stage.

[0014] The path generation and sorting module employs an improved greedy algorithm, introducing a comprehensive cost function:

[0015] ;

[0016] In the formula, Candidate flight segments Overall cost score; This is an unvisited spraying segment currently awaiting evaluation; From the current drone position to the flight segment Minimum entry distance; To complete the voyage The return distance from the current position to the selected take-off and landing point; For the execution of the flight segment The subsequent resource security margin coefficient; , , These are, respectively, the entry distance weight, the return distance weight, and the safety margin weight;

[0017] The dynamic consumption adjustment model established by the material and power estimation module is as follows:

[0018] , ;

[0019] In the formula, The amount of spraying material required to complete the current flight segment; The material consumption coefficient per unit distance indicates the amount of material that the drone needs to spray per unit distance it flies. The horizontal projection length of the flight segment is the path length that the drone needs to spray. Battery power consumption required to complete the current flight segment; The power consumption coefficient per unit horizontal distance reflects the energy efficiency of the drone in horizontal uniform speed flight. The power consumption coefficient per unit altitude reflects the additional energy consumption caused by the drone's vertical climb; The net climb altitude of the flight segment is the difference between the altitude of the endpoint and the altitude of the starting point.

[0020] coefficient , , Adjustments are made in real time based on environmental factors such as wind speed, temperature, and load.

[0021] Furthermore, the UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a CAD drawing parsing module that interacts with the path generation and sorting module.

[0022] The CAD drawing parsing module automatically reads the line type primitives in the DXF file, extracts and standardizes the two-dimensional coordinate sequence of the road spraying segment, generates a structured, non-redundant and geometrically accurate segment dataset, and transmits it to the path generation and sorting module.

[0023] Furthermore, the CAD drawing parsing module automatically parses LINE / LWPOLYLINE primitives in the DXF format file, extracts the key coordinate point sequence of the flight segment, and completes direction standardization, overlap merging, and boundary correction; it also defines coordinate transformation functions. Provides structured input for planar path planning;

[0024] In the formula, A three-dimensional point in the original CAD drawing. or ,Include coordinate; It is a coordinate transformation function used to project or simplify a point in three-dimensional space into two-dimensional plane coordinates; Point The horizontal coordinate on the horizontal plane; Point The vertical coordinate on the horizontal plane; These are the transformed two-dimensional coordinates, representing the position of the point within the working plane; This represents the two-dimensional Euclidean space consisting of all pairs of real numbers, i.e., the Cartesian coordinate system, which is used as input for subsequent path planning algorithms.

[0025] Furthermore, the UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a take-off and landing point management module that interacts with the material and power estimation module and the path generation and sorting module.

[0026] The take-off and landing point management module introduces an accessibility scoring function that integrates distance, time, and resource safety margins. Under the premise of meeting the power and material safety thresholds, it selects the optimal return target from multiple candidate take-off and landing points.

[0027] Furthermore, the accessibility scoring function is:

[0028] ;

[0029] In the formula, Candidate take-off and landing points The overall accessibility score is as follows: the higher the value, the better. Candidate take-off and landing points; This indicates the current location of the drone; To start from the current location of the drone to candidate take-off and landing points The range of flight; The maximum return distance allowed for the mission; To fly from the current location Predicted flight time; The maximum allowable return flight time for the mission; Safety margin coefficient reflects the degree of resource redundancy; , , The dynamically adjustable weight parameters are for the distance, time, and safety margin terms, respectively, to satisfy... This is used to adapt to different task priorities;

[0030] Safety margin coefficient Defined as:

[0031] ;

[0032] In the formula, This indicates the drone's current remaining battery power. To fly to the take-off and landing point Required return battery power; This represents the current remaining amount of material to be sprayed by the drone. To reserve or retain the necessary materials that may be consumed or needed during the return journey;

[0033] Only for those who meet the requirements and The take-off and landing points are scored and optimized.

[0034] Furthermore, the UAV spraying system for snow accumulation identification and dynamic path planning on grade-separated roads also includes a supply determination module that interacts with the material and power estimation module and the take-off and landing point management module.

[0035] The resupply determination module constructs a dynamic threshold model based on remaining resources, return distance, and environmental factors to determine the resupply needs in real time, and records the triggering conditions and resupply types, supporting task traceability and system optimization.

[0036] Furthermore, the UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a dynamic task splitting module that interacts with the path generation and sorting module.

[0037] The dynamic task splitting module is based on real-time resource assessment and safety margin judgment. It automatically interrupts and divides the remaining tasks during the execution of a flight segment, ensuring that the operation does not fail midway due to power or material depletion, and supports the traceability and reconstruction of task status.

[0038] Furthermore, this drone spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes an output and visualization module that interacts with each module;

[0039] The output and visualization module displays the entire mission process through a graphical interface and generates structured reports containing data on flight, spraying, and resource consumption. It supports export in multiple formats for archiving and analysis.

[0040] The beneficial effects of this invention are:

[0041] The invention, "Unmanned Aerial Vehicle Spraying System for Snow Recognition and Dynamic Path Planning on Interchanges," demonstrates significant technical advantages and practical value in snow removal operations on complex interchanges through multi-module collaboration and intelligent closed-loop design. Specific beneficial effects are as follows:

[0042] 1. Achieve high-precision, low-latency real-time snow cover perception: Employ a lightweight FLSD model that integrates depthwise separable convolution, CSP structure, and multi-scale attention mechanism to achieve an inference speed of 20ms / frame and high detection accuracy at the edge of the drone, effectively addressing complex road scenarios such as strong reflections, shadows, and background interference, and providing a reliable perception foundation for "on-demand spraying".

[0043] 2. Breaking through the limitations of traditional static path planning, achieving dynamic intelligent scheduling: Based on snow distribution, resource status and safety margin, an improved greedy algorithm is used to dynamically generate spraying paths, comprehensively optimizing entry distance, return feasibility and resource redundancy, avoiding the operational risk of "being close but unable to return", and significantly improving path rationality and task continuity.

[0044] 3. Improve resource utilization efficiency and reduce operation and maintenance costs: Accurately predict material and power demand through dynamic consumption models, and combine dynamic task splitting and replenishment judgment mechanisms to avoid ineffective flight and excessive spraying, thus significantly reducing material and energy waste.

[0045] 4. Ensure operational safety and mission reliability: Introduce a take-off and landing point reachability scoring function and a safety margin threshold screening mechanism to ensure that the UAV always operates within the safety boundary; when the mission is interrupted, it will automatically split and record the status, support breakpoint resume flight, and eliminate the risk of mid-flight failure.

[0046] 5. Achieve full-process automation and visualization management: From automatic parsing of flight segments from CAD drawings to snow cover recognition, route planning, and execution feedback, no manual intervention is required throughout the entire process; the output module provides a graphical task view and structured reports, supports CSV / PDF export, and facilitates scheduling supervision, performance evaluation, and strategy optimization.

[0047] 6. Particularly suitable for complex interchange scenarios, filling a technological gap: It provides flexible, safe and efficient snow removal solutions for high-risk areas such as elevated bridges, multiple ramps, and sharp bends that are difficult for large equipment to cover, significantly improving the winter traffic safety of urban expressway networks.

[0048] 7. Possesses good scalability and engineering implementation potential: The system architecture supports expansion to advanced functions such as multi-machine collaborative operation, real-time obstacle avoidance, and weather linkage early warning. Furthermore, the model and algorithm have been verified on embedded platforms such as Jetson Xavier NX and are ready for practical deployment.

[0049] In summary, this system not only solves the core pain points of existing snow removal technologies in terms of perception, planning, and resource management, but also promotes the upgrading of municipal winter maintenance towards intelligence, precision, and greenness, demonstrating outstanding technological progress and broad application prospects. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0051] Figure 1 A schematic diagram of the composition of a drone spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads;

[0052] Figure 2 This is a schematic diagram of the overall architecture of the road snow accumulation recognition module;

[0053] Figure 3 Image showing the effect of identifying snow accumulation on roads;

[0054] Figure 4 A schematic diagram of the overall highway interchange and air route CAD drawing;

[0055] Figure 5 This is a visual diagram of the flight path for mission segment 10. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0057] This embodiment provides a drone spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads, aiming to solve the problems of perception lag, rigid path, resource waste and high operational risks faced by traditional road snow removal methods on complex grade-separated roads such as viaducts, multiple ramps and sharp bends.

[0058] like Figure 1 As shown, this UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads includes: a road snow accumulation recognition module, a path generation and sorting module, and a material and power estimation module. The specific details of each module are as follows:

[0059] Road snow accumulation recognition module:

[0060] The road snow recognition module is deployed on the UAV's onboard edge computing unit, such as NVIDIA Jetson XavierNX or a domestic embedded SoC. It is used to detect snow areas in real-time video streams collected by onboard visible light and infrared cameras before or during spraying tasks, and output structured perception results for the path generation and sorting module to dynamically adjust the operation strategy.

[0061] like Figure 2 As shown, the road snow recognition module, based on the improved YOLOv10 target detection network, constructs a fast and lightweight snow surface detection model, FLSD, the specific implementation of which is as follows:

[0062] Step 1. Lightweight Network Structure: Depthwise Separable Convolution is introduced into the backbone network, decomposing standard convolution into channel-wise convolution and 1×1 pointwise convolution, significantly reducing computational cost and parameter count. Simultaneously, a Cross-Stage Partial Connection (CSP) structure is adopted, dividing the input feature channels into preserved and transformed paths, reducing redundant information transmission through cross-layer fusion. After optimization, the model size is reduced to approximately 60% of the original YOLOv10 backbone, inference speed is improved by 25%, while mAP accuracy remains largely unchanged, effectively adapting to the computational constraints of UAV embedded platforms.

[0063] Step 2. Feature Extraction Enhancement: To address the challenges of identifying snow accumulation on roads under strong light reflection, shadow occlusion, and complex backgrounds such as tire tracks and vegetation, a multi-scale attention mechanism is introduced in the feature fusion stage. This mechanism includes: a spatial attention module, which weights each spatial location of the feature map to enhance the boundary discrimination ability between high-reflectivity areas and shadow areas; a channel attention module, which adaptively adjusts channel weights to highlight snow surface texture and edge features while suppressing irrelevant interference; and a multi-scale feature fusion structure, which aggregates high, medium, and low resolution features to accommodate both large-area snow cover and the detection of small residual patches.

[0064] like Figure 3 As shown, this system underwent multiple comparative experiments in real-world snow-covered road scenarios, focusing on verifying the robustness of the FLSD model under complex lighting conditions such as backlighting, specular reflection from snow, and high dynamic range backgrounds. The green rectangles in the image represent the boundaries of snow-covered areas successfully identified by the model. Despite strong sunlight reflection and shadow interference in the image, the model was still able to accurately define the snow cover area, unaffected by tire tracks or roadside vegetation.

[0065] The experiment was based on a self-built "Snow Accumulation Dataset for Interchanges," covering various extreme lighting scenarios such as morning backlight, afternoon strong reflection, twilight low illumination, and clear skies after snowfall. Compared with the original YOLOv10 model, the FLSD model, after introducing a multi-scale attention mechanism, improved its mean accuracy (mAP@0.5) from 78.2% to 82.9% under the aforementioned complex conditions, a relative improvement of approximately 4.7 percentage points. Simultaneously, the false positive rate decreased by 12.3%, and the false negative rate decreased by 8.6%, indicating that this mechanism effectively enhanced the model's ability to distinguish between bright reflective areas and blurred texture regions.

[0066] The results fully demonstrate that the proposed multi-scale attention fusion structure can significantly improve the stability of UAV snow perception in real complex environments, providing a reliable input basis for subsequent dynamic path planning.

[0067] Step 3. Edge Inference Optimization: To meet real-time requirements, deployment-level optimization is performed on the trained model: Structured pruning is adopted, and redundant convolutional kernels are removed based on channel importance scoring, reducing the computational cost to 65% of the original while preserving key feature pathways; INT8 weight quantization is implemented to convert the FP32 model to INT8 format, improving inference speed by 1.8-2.2 times on ARM NPU or GPU while keeping accuracy loss within 1%; Operator fusion and memory access optimization are performed in combination with hardware characteristics to achieve pipelined parallel processing of video decoding, preprocessing, and inference, with end-to-end latency stabilized at 20ms / frame and 50FPS, meeting the requirements of real-time operations.

[0068] During mission execution, this module outputs structured snow cover information per frame, including: coordinates of the snow cover area bounding box, coverage ratio estimated based on pixel percentage, and confidence score. This data is transmitted in real-time to the path generation and sorting module, which dynamically adjusts the spraying path sequence, flight speed, or localized operation mode, thereby significantly improving snow removal efficiency and resource utilization while ensuring operational safety.

[0069] Path generation and sorting module:

[0070] The path generation and sorting module is used within each spraying task to dynamically generate efficient, safe, and continuously executable spraying paths based on the UAV's current starting position and the structured flight segment set provided by the system. This module is deployed in the UAV's onboard mission management unit or ground control station and can combine real-time sensing information and resource status to achieve flexible task scheduling.

[0071] The "structured flight segment set" is provided in advance by the system. Its source can be manually delineated work areas, road network data imported from GIS maps, or road sections that can be sprayed defined by other means. Each flight segment is represented in the form of a two-dimensional coordinate point sequence, which constitutes the smallest work unit that the UAV can perform.

[0072] The specific implementation process is as follows:

[0073] Step 1. Input Data Preparation: The module receives the following key inputs: a set of flight segments to be sprayed, each segment defined by a start point, an end point, and intermediate waypoints; real-time snow detection results from the road snow recognition module, used to filter the flight segments to be actually sprayed, such as only spraying segments with a snow coverage ratio higher than a threshold; and the current real-time location of the drone. Preferred take-off and landing points Remaining battery power With remaining materials .

[0074] Step 2. Path Strategy Selection: The system supports multiple path planning strategies, including: "Start Point Priority" mode: Prioritizes the selection of the flight segment closest to the current UAV position to start the operation, which is suitable for rapid response scenarios; "Sequence Maintenance" mode: Executes according to the preset logical order of flight segments, such as along the ramp, which is suitable for curves or slopes where the continuity of the operation is required; The strategy can be specified by the operator or automatically switched by the system according to the urgency of the task.

[0075] Step 3. Execution of the improved greedy path sorting algorithm: The module uses an improved greedy algorithm, combined with resource safety constraints, to iteratively generate the segment access sequence according to the following steps:

[0076] First, initialize the set of unvisited flight segments that need to be sprayed. ;

[0077] Then, for each candidate flight segment Calculate its minimum entry distance This refers to the shortest flight distance from the current location to any entry point of this flight segment.

[0078] Then, the distance to the preferred take-off and landing point after executing this flight segment is calculated. return distance ;

[0079] Then, based on the material and power estimation model, the execution is predicted. The remaining resources after calculation, and the safety margin factor. :

[0080] ;

[0081] In the formula, , The remaining electricity and materials after execution; , Resources needed for the return flight;

[0082] Then, construct the comprehensive cost function:

[0083] ;

[0084] In the formula, Candidate flight segments Overall cost score; This is an unvisited spraying segment currently awaiting evaluation; From the current drone position to the flight segment Minimum entry distance; To complete the voyage The return distance from the current position to the selected take-off and landing point; For the execution of the flight segment The subsequent resource security margin coefficient; , , These are, respectively, the entry distance weight, the return distance weight, and the safety margin weight. , , >0 indicates a dynamically adjustable weighting coefficient; for example, it will automatically increase when the battery level is below 30%. and Prioritizing the safety of the return flight;

[0085] Then, select The segment with the smallest value is selected as the next task segment, added to the path sequence, and from... Remove from;

[0086] Finally, update the current starting point to the end point of the flight segment, and repeat the above process until all flight segments that need to be sprayed have been processed or the resupply conditions are triggered.

[0087] Step 4. Local Optimization and Continuity Guarantee: Based on the greedy main line, the module can perform lightweight optimizations such as 2-opt swaps on local subsequences of the generated path, such as three consecutive flight segments; thereby reducing unnecessary turnarounds and improving path smoothness and operational efficiency.

[0088] Through the above mechanism, the path generation and sorting module realizes dynamic path planning driven by perception, constrained by resources, and prioritizing safety based on the input of general structured flight segments. This effectively avoids the problems of "being close to the destination but unable to return" or "running out of resources and stopping midway" in traditional methods, and significantly improves the reliability and intelligence level of snow removal operations on complex grade-separated roads.

[0089] Materials and power estimation module:

[0090] The material and power estimation module is deployed in the UAV onboard mission management system. It is used to perform high-precision dynamic estimation of the spraying materials and battery power required for the current flight segment or mission subset before and during mission execution, providing key resource constraints for dynamic mission splitting, path planning and resupply determination.

[0091] The inputs to the materials and power estimation module include: the geometric parameters of the flight segment and the horizontal projected length. and net climb height The drone's current status parameters include: payload, flight speed, and spraying rate; real-time environmental data obtained through onboard sensors or meteorological APIs, such as wind speed, temperature, and air density.

[0092] The materials and power estimation module has a built-in dynamic resource consumption model, as detailed below:

[0093] Material consumption model: Spraying material demand Calculate using the following formula:

[0094] ;

[0095] in, The amount of spraying material required to complete the current flight segment; The horizontal projection length of the flight segment is the path length that the drone needs to spray. The material consumption coefficient per unit distance represents the amount of material required to be sprayed by the drone per unit distance it flies. For example, if the nozzle flow rate is 0.5 L / min and the flight speed is 5 m / s, then... =0.0017L / m.

[0096] In actual operation, It can dynamically adjust according to environmental and task requirements: when a high snow cover ratio is detected or the temperature is below -5℃, it automatically increases. To increase the concentration of de-icing agent; when encountering crosswinds or headwinds, to compensate for drift loss, the concentration should be appropriately increased. The system supports presets by region. Values ​​are used to achieve differentiated and precise spraying.

[0097] Power consumption model: Battery power demand Calculate using the following formula:

[0098] ;

[0099] in, Battery power consumption required to complete the current flight segment; The power consumption coefficient per unit horizontal distance reflects the energy efficiency of the drone in horizontal uniform speed flight. The power consumption coefficient per unit altitude reflects the additional energy consumption caused by the drone's vertical climb; This is the net climb altitude of the flight segment, which is the difference between the altitude of the endpoint and the altitude of the starting point.

[0100] coefficient and It is not a fixed constant, but is adjusted in real time based on the following factors: low temperatures increase battery internal resistance and decrease efficiency. , Automatic adjustment, such as increasing by 15% at 0°C compared to 25°C. Headwinds increase drag. The weight increases linearly in segments according to wind speed. As materials are consumed, the overall weight of the drone decreases. , Gradually reduce the power level. During the climb phase, activate high-power mode. Dynamic enhancement; additional energy consumption is introduced when hovering or turning at low speed.

[0101] Assuming a certain flight segment length =300m, elevation gain =15m, current temperature -3℃, wind speed (headwind) 4m / s, remaining load 12kg. The system queries the correction factor for the current environment: =0.0021L / m, =0.85Wh / m, =12.3Wh / m. Therefore: =0.0021×300=0.63L,

[0102] =0.85×300+12.3×15=255+184.5=439.5Wh. This result is output in real time to the dynamic task splitting module and the replenishment judgment module to determine whether the safety margin condition is met.

[0103] Through the aforementioned dynamic modeling and real-time correction mechanisms, the material and power estimation module significantly improves the accuracy and environmental adaptability of resource estimation, effectively supporting the system's intelligent decision-making and safe operation in complex grade-separated road scenarios.

[0104] Based on the aforementioned road snow accumulation recognition module, path generation and sorting module, and material and power estimation module, a drone-based intelligent spraying system integrating "perception → decision-making → resource constraints" is constructed to achieve the winter road maintenance goals of on-demand response, precise snow removal, and optimal resource utilization.

[0105] Furthermore, as a preferred technical solution in this embodiment, such as Figure 1As shown, the UAV spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads also includes a CAD drawing parsing module. The CAD drawing parsing module interacts with the path generation and sorting module to automatically convert road design CAD drawings provided by municipal or transportation departments into structured, executable spraying segment data, providing high-precision geometric input for subsequent path planning.

[0106] like Figure 4 As shown, the original CAD drawings processed by the CAD drawing parsing module represent the overall road network structure of a highway interchange, including multiple ramps, main roads, and intersections. The graphic elements are represented by colored lines indicating different functional road sections, such as green for main roads, purple for ramps, and yellow for connecting passages. By parsing such drawings, the system automatically extracts the boundaries of spraying operations and performs standardization processing.

[0107] The specific implementation process is as follows:

[0108] Step 1. Input and Element Recognition: The module receives a standard DXF format CAD drawing file, which typically contains vector elements of road structures such as overpasses, ramps, and main roads. The system traverses all entities in the DXF file, focusing on identifying line elements used to define sprayable areas, including straight lines (LINE) and lightweight polylines (LWPOLYLINE), and filtering out text, annotations, fills, and other non-operation-related elements.

[0109] Step 2. Segment Coordinate Extraction and Preprocessing: For each valid primitive, extract its vertex coordinate sequence. For LWPOLYLINE, obtain all its control points; for LINE, treat it as a segment consisting of two endpoints. All raw coordinates are stored in three-dimensional vector form, denoted as... ;wherein is a three-dimensional spatial point in the original CAD drawing; This is elevation information.

[0110] Step 3. Coordinate Transformation and 2D Projection: Since the drone spraying path planning is mainly performed on the horizontal plane, the system defines a coordinate transformation function:

[0111] ;

[0112] In the formula, It is a coordinate transformation function used to project or simplify a point in three-dimensional space into two-dimensional plane coordinates; Point The horizontal coordinate on the horizontal plane; Point The vertical coordinate on the horizontal plane; These are the transformed two-dimensional coordinates, representing the position of the point within the working plane; This represents the two-dimensional Euclidean space consisting of all pairs of real numbers, i.e., the Cartesian coordinate system, which is used as input for subsequent path planning algorithms.

[0113] Step 4. Geometry Post-processing and Data Optimization: To ensure the accuracy and feasibility of the flight segment data, the module performs the following geometry optimization operations:

[0114] Directional standardization: Unify the direction of all flight segments, such as sorting them according to traffic flow direction or clockwise, to avoid directional conflicts during route planning; Overlapping flight segment merging: Detect and merge adjacent line segments that overlap in space or are at similar heights to eliminate redundancy and reduce task fragmentation; Automatic correction of boundary flight segments: Automatically repair breakpoints, misalignments, or minor gaps caused by drawing errors using interpolation or nearest neighbor connection strategies to ensure continuous and closed flight segments.

[0115] Step 5. Structured Flight Segment Dataset Output: Finally, the module generates a structured flight segment dataset, where each flight segment is represented as an ordered list of two-dimensional coordinate points:

[0116] ;

[0117] The dataset is transmitted to the path generation and sorting module through an internal interface, serving as the basic task space for dynamic path planning and snow cover response spraying scheduling.

[0118] Through the above processing, the CAD drawing parsing module realizes the automated bridging from engineering design drawings to intelligent UAV mission instructions, which significantly improves the deployment efficiency and operation accuracy of the system in complex interchange road scenarios, while avoiding the subjective errors and time costs of manually drawing flight paths.

[0119] Furthermore, as a preferred technical solution in this embodiment, such as Figure 1 As shown, this UAV spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads also includes a take-off and landing point management module. This module interacts closely with the material and power estimation module and the path generation and sorting module to intelligently select the optimal return target in multi-take-off and landing point deployment scenarios, ensuring that the UAV always has a safe and efficient return capability during complex operations.

[0120] When the system has multiple candidate take-off and landing points preset, such as emergency parking aprons, service areas or designated safe areas around overpasses, the take-off and landing point management module evaluates the accessibility of each candidate point in real time during the task execution process and makes dynamic decisions based on resource status.

[0121] The specific implementation process is as follows:

[0122] Step 1: Initial Feasibility Screening: For each candidate take-off and landing point The takeoff and landing point management module first calls the materials and power estimation module to obtain the resources required for the return flight, including:

[0123] Fly to Required return battery power. : The remaining amount of spray material that needs to be retained or consumed during the return flight; in some scenarios, no material is consumed during the return flight, so it can be set to 0, but for safety reasons, a minimum reserve is usually retained.

[0124] Simultaneously obtain the current status of the drone:

[0125] Remaining battery power; Remaining material quantity. Introducing a safety margin threshold. , A landing site is considered a feasible candidate only if the following two conditions are met:

[0126] and ;

[0127] Take-off and landing points that fail the screening are directly excluded to avoid the risk of being forced to return at the last minute.

[0128] Step 2: Comprehensive Accessibility Score: For all feasible candidate take-off and landing points, calculate their accessibility score function.

[0129] ;

[0130] In the formula, Candidate take-off and landing points The overall accessibility score is as follows: the higher the value, the better. Candidate take-off and landing points; This indicates the current location of the drone; To start from the current location of the drone to candidate take-off and landing points The range of flight; The maximum return distance allowed for the mission; To fly from the current location Predicted flight time; The maximum allowable return flight time for the mission; Safety margin coefficient reflects the degree of resource redundancy; safety margin coefficient Defined as:

[0131] ;

[0132] In the formula, This indicates the drone's current remaining battery power. To fly to the take-off and landing point Required return battery power; This represents the current remaining amount of material to be sprayed by the drone. To reserve or retain the necessary materials that may be consumed or needed during the return journey;

[0133] , , The dynamically adjustable weight parameters are for the distance, time, and safety margin terms, respectively, to satisfy... This is used to adapt to different task priorities, for example:

[0134] During an emergency return, increase Prioritize the takeoff and landing points with the shortest travel time;

[0135] During routine operations, increase Prioritize ensuring resource security.

[0136] Step 3: Optimal Takeoff and Landing Point Selection and Output: The module selects and outputs all feasible candidate points according to... Sort by highest to lowest and select the highest-rated. The coordinates of this point are used as the current optimal return target and are transmitted in real time to the path generation and sorting module for calculating the return distance. And update the comprehensive cost function.

[0137] Through the above mechanism, the take-off and landing point management module achieves intelligent optimization of multiple take-off and landing points while ensuring safety margins. This significantly improves the robustness and mission reliability of UAVs during long-term operations on complex overpasses and effectively avoids the risk of crashes or mission interruptions caused by blindly returning to base.

[0138] Furthermore, as a preferred technical solution in this embodiment, such as Figure 1 As shown, the UAV spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads also includes a resupply determination module. This module interacts in real time with the material and power estimation module and the take-off and landing point management module to dynamically assess whether the current operation needs to be interrupted and the system returned for resupply during mission execution, thereby ensuring the system's continuous, safe, and efficient operation.

[0139] The specific implementation process is as follows:

[0140] Step 1. Dynamic Threshold Model Construction: The replenishment determination module constructs an adaptive threshold determination model based on multi-source inputs. The core inputs include: the drone's current remaining battery power. and remaining amount of sprayed material From the materials and power estimation module; flying to the current preferred takeoff and landing point. Required return battery power Material allowance Real-time environmental factors, such as wind speed, temperature, and air density, are obtained through airborne sensors or meteorological interfaces; mission phase information, such as whether it is in a high-priority ramp operation.

[0141] The system dynamically adjusts the replenishment trigger threshold based on the above factors, rather than using a fixed percentage. For example, when the wind speed increases or the temperature decreases, the energy consumption for returning to base increases, and the system automatically raises the power warning threshold. In remote operating areas far from the take-off and landing points, even if there are still plenty of resources, replenishment may be triggered earlier due to the long return distance. If it is detected that the next flight segment has severe snow accumulation and requires high-concentration spraying, the material threshold will be tightened accordingly.

[0142] Step 2. Supply condition judgment logic: The module continuously monitors whether any of the following conditions are met:

[0143] ;

[0144] ;

[0145] in , To dynamically adjust the safety margin factor according to the environment and task, it is usually , Once any of the conditions are met, the return-to-base resupply command will be triggered immediately.

[0146] Step 3. Supply Event Recording and Type Labeling: When supply is triggered, the module automatically records a complete event log, including: trigger time; current task location; and remaining resource status. , The return destination point is from the takeoff and landing point management module.

[0147] Supply type tags include, but are not limited to: "Low Power": triggered due to power falling below the dynamic threshold; "Materials Depleted": triggered due to insufficient snow melting agent reserves; "Task Segment Completed + Resource Critical": the current subtask has ended and resources are nearing the safety boundary; "Sudden Environmental Change": such as a sudden surge in energy consumption caused by strong winds. Logs are stored in a structured format, such as JSON or database records, and synchronized to the output and visualization module for generating task reports.

[0148] Step 4. Coordination with Task Scheduling: The resupply determination result is notified to the dynamic task splitting module in real time. Based on this, the module divides the incomplete flight segment into new tasks, ensuring that the operation can be interrupted and resumed. At the same time, after receiving the resupply instruction, the path generation module immediately plans the return route, prioritizing a safe landing.

[0149] Through the above mechanism, the replenishment determination module has realized the transformation from "passive low-battery return" to "active predictive replenishment". This not only avoids the risk of failure during the mission, but also provides data support for the long-term operation of the system, and supports the continuous optimization of consumption model, path strategy and take-off and landing point layout.

[0150] Furthermore, as a preferred technical solution in this embodiment, such as Figure 1 As shown, the UAV spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads also includes a dynamic task splitting module. This module interacts closely with the path generation and sorting module and relies on resource prediction data provided by the material and power estimation module to achieve intelligent and safe operation interruption and task reconfiguration during task execution, ensuring that the UAV is not forced to stop midway due to power or depletion of spraying materials.

[0151] The specific implementation process is as follows:

[0152] Step 1. Triggering Timing and Evaluation Mechanism: After each flight segment is completed, i.e., when the UAV finishes a spraying operation and reaches the end of that segment, the system immediately initiates the mission continuity evaluation process. The module calls the material and power estimation module to obtain the next flight segment to be accessed. Required resource consumption forecast and And combined with the current remaining resources , Make a judgment.

[0153] Step 2. Safety Margin Judgment Logic: Introduce a dynamic safety margin threshold. , If any of the following conditions are met, the next flight segment will be deemed unsafe to complete, triggering a mission split:

[0154] ;

[0155] ;

[0156] This mechanism ensures that even if there are sudden environmental changes, such as increased headwinds or minor errors in the model, the system still has enough redundancy to complete the current flight segment and return safely.

[0157] Step 3. Mission Interruption and Split Operation: Interrupt the current mission flow and send a "Terminate Spraying, Prepare for Return" command to the flight control system; aggregate the remaining unexecuted segments, including... The original task and its subsequent segments are encapsulated as a new subtask and assigned a unique task ID; the task status database is updated, marking the original task as "partially completed" and the new subtask as "pending scheduling"; the path generation and sorting module is notified to replan the spraying path of the subtask based on the new starting point before the next takeoff.

[0158] Step 4. Task Status Log Recording and Traceability: The module synchronously generates a structured task segment log, recording the following key information: original task ID and new subtask ID; time of occurrence, geographical location, and list of completed segments; reason for segmentation, such as "insufficient power" or "material criticality"; remaining resource status at the time of segmentation and prediction basis. The log supports complete traceability of task status, visual playback, and fault debriefing, and is used for task reconstruction when the system resumes operations, i.e., seamless continuation from the breakpoint, avoiding repeated spraying or missed areas.

[0159] Step 5. Coordinate with the replenishment process: The dynamic task splitting module and the replenishment judgment module work together: If the splitting is caused by insufficient resources, the system will automatically enter the replenishment process; after the replenishment is completed, the scheduler will prioritize loading the subtask to realize the "interruption-replenishment-continued flight" closed loop.

[0160] Through the above mechanism, the dynamic task splitting module significantly improves the robustness and reliability of the system in complex and long-term operation scenarios, and truly realizes the intelligent operation capability of "task can be interrupted, state can be saved, and operation can be continued", effectively solving the technical bottleneck of traditional drone spraying systems that cause task failure due to resource depletion.

[0161] Furthermore, as a preferred technical solution in this embodiment, such as Figure 1 As shown, this UAV spraying system for snow accumulation recognition and dynamic path planning on grade-separated roads further includes an output and visualization module. This module serves as the system's human-computer interaction and data archiving hub, interacting in real-time with multiple functional units such as the road snow accumulation recognition module, CAD drawing parsing module, path generation and sorting module, dynamic task splitting module, supply determination module, and material and power estimation module. It aggregates data throughout the entire task lifecycle, providing intuitive graphical displays and structured report outputs.

[0162] The specific implementation process is as follows:

[0163] Step 1. Multi-source data aggregation: During task execution, the module continuously receives key data streams from various sub-modules, including: original CAD drawings and the parsed structured flight segment set from the CAD drawing parsing module; real-time snow cover detection results, such as snow area heat maps and coverage ratios, from the road snow cover recognition module; dynamically generated spraying paths, flight segment access order, and return trajectories from the path generation and sorting module; task splitting records and resupply event logs, including trigger time, location, and cause from the dynamic task splitting module and resupply determination module; and flight distance, flight time, spraying area, and material usage for each flight segment from the material and power estimation module and flight logs. Power consumption Resource data, etc.

[0164] Step 2. Graphical Interface Display: Based on OpenCV, Matplotlib, or a lightweight GUI framework, modules construct a unified visualization interface to dynamically present the following content:

[0165] Background map overlay: Optionally load satellite imagery or CAD base map;

[0166] Segment distribution: Completed, pending, and skipped segments without snow are marked with different colors;

[0167] Drone trajectory: Real-time plotting of flight path, marking current position and heading;

[0168] Key node marking: Use icons to mark take-off and landing points, supply points, and mission split locations;

[0169] Resource status curve: Simultaneously displays the changing trends of electricity and materials over time / distance;

[0170] Snow response heatmap: Mapping the identification results onto the road surface to intuitively reflect the effect of "on-demand spraying".

[0171] All graphic elements support scaling, panning, and layer switching, facilitating real-time monitoring and post-event review by dispatchers.

[0172] Step 3. Structured Task Report Generation: After the task is completed, the module automatically generates a detailed task report, including but not limited to the following:

[0173] Basic task information: Task ID, Execution Date, Job Area;

[0174] Segment execution details: segment number, start / end coordinates, length, flight time, spraying area, material usage, and power consumption;

[0175] Resource summary: total flight distance, total spray volume, total power consumption, and material utilization rate;

[0176] Abnormal event log: Number of resupply attempts, number of task splits, triggering reasons, and handling results;

[0177] Performance metrics include total path length, number of return trips, and operational continuity score.

[0178] Step 4. Multi-format export and archiving support: The report supports exporting to multiple standard formats:

[0179] CSV format: easy to import into Excel or databases for statistical analysis;

[0180] PDF format: Generate formal documents with charts for archiving, reporting, or auditing;

[0181] JSON / XML format: for use by the upper-level smart municipal platform to achieve data exchange.

[0182] All exported files are automatically named by task ID and stored in a specified directory locally or in the cloud. They can be searched by time, region, drone number, and other dimensions.

[0183] Through the above design, the output and visualization module not only improves the operability and transparency of the system, but also provides a high-quality data foundation for operation and maintenance optimization, strategy iteration and performance evaluation. It is an important component of the "intelligent perception-autonomous decision-making-closed-loop feedback" technical system.

[0184] In addition, to verify the UAV spraying system for snow accumulation recognition and dynamic path planning at interchanges, this embodiment also conducted the following practical verifications:

[0185] like Figure 4 As shown, Figure 4 This is an analytical rendering of the overall CAD drawing of the highway interchange and its flight segments. The system successfully extracted and standardized the coordinate sequences of multiple ramps, main roads, and connecting passages from the original DXF drawings, completing direction unification, overlap merging, and boundary correction. In the drawing, colored lines represent different functional areas; for example, green represents main roads, purple represents ramps, gray dashed lines represent auxiliary connecting paths automatically generated by the system, and white "+" signs mark candidate takeoff and landing points. Manual comparison showed that the geometric accuracy error of the flight segments is less than 0.5 meters, meeting the requirements for subsequent path planning and spraying operations.

[0186] like Figure 5 As shown, Figure 5 This is a visualization of the flight path for Task Segment 10. This task comprises multiple consecutive segments. Based on the current UAV location and snow distribution information, the system uses an improved greedy algorithm to generate the optimal access order. In the diagram, the solid blue line represents the actual flight trajectory, the dashed line represents the return path, and the red dots "Start Point 1" and "End Point 2" indicate the task start and end positions, respectively. The data displayed in the diagram shows: total flight distance 1183.76 meters, total time 4.47 minutes, and material consumption 76.65 kg. The path planning effectively avoids repeated backtracking and dynamically inserts return nodes based on resource status during execution, ensuring task integrity.

[0187] Through the above-mentioned field deployment and data verification, this system has achieved the following results in complex interchange scenarios:

[0188] Automated conversion of CAD drawings into structured flight segments; closed-loop response for dynamic path adjustment based on snow cover perception; intelligent decision-making for task splitting and resupply based on resource constraints; and full support for graphical visualization and data archiving throughout the entire mission process.

[0189] Experimental results show that this system can significantly improve the efficiency of snow removal operations, reduce the path length by about 18% compared with traditional fixed routes, reduce material waste by more than 15%, and no power outages or crashes occurred midway. Therefore, this system has good engineering practical value and promotion prospects.

[0190] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A drone spraying system for snow accumulation recognition and dynamic path planning at grade-separated roads, characterized in that: The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads includes: a road snow accumulation recognition module, a path generation and sorting module, and a material and power estimation module; The road snow recognition module is based on a lightweight FLSD model, which enables low-latency and high-precision real-time snow detection under complex lighting conditions at the edge of the drone, and outputs the real-time snow detection results to the path generation and sorting module. The path generation and sorting module uses a comprehensive cost function that integrates entry distance, return distance and resource safety margin, based on the current starting position and the set of flight segments, and dynamically generates continuously executed spraying paths through an improved greedy algorithm. The material and power estimation module, based on flight segment geometric parameters and environmental factors, uses a dynamically adjusted consumption model to accurately predict the demand for spraying materials and battery power in real time, providing a basis for resource constraints for mission scheduling and resupply decisions. The road snow recognition module is based on the improved YOLOv10 to build a lightweight model FLSD. The lightweight model FLSD introduces a depthwise separable convolution + CSP structure in the backbone network. In the feature extraction stage, it introduces an attention mechanism that integrates spatial attention and channel attention with multi-scale feature fusion. In the edge inference stage, it uses structured pruning, INT8 weight quantization, and operator fusion for real-time edge inference. The path generation and sorting module employs an improved greedy algorithm, introducing a comprehensive cost function: ； In the formula, Candidate flight segments Overall cost score; This is an unvisited spraying segment currently awaiting evaluation; From the current drone position to the flight segment Minimum entry distance; To complete the voyage The return distance from the current position to the selected take-off and landing point; For the execution of the flight segment The subsequent resource security margin coefficient; , , These are, respectively, the entry distance weight, the return distance weight, and the safety margin weight; The dynamic consumption adjustment model established by the material and power estimation module is as follows: , ； In the formula, The amount of spraying material required to complete the current flight segment; The material consumption coefficient per unit distance indicates the amount of material that the drone needs to spray per unit distance it flies. The horizontal projection length of the flight segment is the path length that the drone needs to spray. Battery power consumption required to complete the current flight segment; The power consumption coefficient per unit horizontal distance reflects the energy efficiency of the drone in horizontal uniform speed flight. The power consumption coefficient per unit altitude reflects the additional energy consumption caused by the drone's vertical climb; The net climb altitude of the flight segment is the difference between the altitude of the endpoint and the altitude of the starting point. coefficient , , Adjustments are made in real time based on environmental factors such as wind speed, temperature, and load.

2. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 1, characterized in that: The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a CAD drawing parsing module, which interacts with the path generation and sorting module. The CAD drawing parsing module automatically reads the line type primitives in the DXF file, extracts and standardizes the two-dimensional coordinate sequence of the road spraying segment, generates a structured, non-redundant and geometrically accurate segment dataset, and transmits it to the path generation and sorting module.

3. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 2, characterized in that: The CAD drawing parsing module automatically parses LINE / LWPOLYLINE primitives in DXF format files, extracts key coordinate point sequences for flight segments, and performs direction standardization, overlap merging, and boundary correction; it also defines coordinate transformation functions. Provides structured input for planar path planning; In the formula, A three-dimensional point in the original CAD drawing. or ,Include coordinate; It is a coordinate transformation function used to project or simplify a point in three-dimensional space into two-dimensional plane coordinates; Point The horizontal coordinate on the horizontal plane; Point The vertical coordinate on the horizontal plane; These are the transformed two-dimensional coordinates, representing the position of the point within the working plane; This represents the two-dimensional Euclidean space consisting of all pairs of real numbers, i.e., the Cartesian coordinate system, which is used as input for subsequent path planning algorithms.

4. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 1, characterized in that: The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a take-off and landing point management module that interacts with the material and power estimation module and the path generation and sorting module. The take-off and landing point management module introduces an accessibility scoring function that integrates distance, time, and resource safety margin. Under the premise of meeting the power and material safety thresholds, it selects the candidate take-off and landing point with the highest accessibility score from multiple candidate take-off and landing points as the return target.

5. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 4, characterized in that: The accessibility scoring function is: ； In the formula, Candidate take-off and landing points The overall accessibility score is as follows: the higher the value, the better. Candidate take-off and landing points; This indicates the current location of the drone; To start from the current location of the drone to candidate take-off and landing points The range of flight; The maximum return distance allowed for the mission; To fly from the current location Predicted flight time; The maximum allowable return flight time for the mission; Safety margin coefficient reflects the degree of resource redundancy; , , The dynamically adjustable weight parameters are for the distance, time, and safety margin terms, respectively, to satisfy... This is used to adapt to different task priorities; Safety margin coefficient Defined as: ； In the formula, This indicates the drone's current remaining battery power. To fly to the take-off and landing point Required return battery power; This represents the current remaining amount of material to be sprayed by the drone. To reserve or retain the necessary materials that may be consumed or needed during the return journey; Only for those who meet the requirements and The candidate take-off and landing points are scored for accessibility, and the candidate take-off and landing point with the highest accessibility score is determined as the return target.

6. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 4, characterized in that: The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a resupply determination module that interacts with the material and power estimation module and the take-off and landing point management module. The resupply determination module constructs a dynamic threshold model based on remaining resources, return distance, and environmental factors to determine the resupply needs in real time, and records the triggering conditions and resupply types, supporting task traceability and system optimization.

7. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 1, characterized in that: The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes a dynamic task splitting module that interacts with the path generation and sorting module. The dynamic task splitting module is based on real-time resource assessment and safety margin judgment. It automatically interrupts and divides the remaining tasks during the execution of a flight segment, ensuring that the operation does not fail midway due to power or material depletion, and supports the traceability and reconstruction of task status.

8. The UAV spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads according to claim 1, characterized in that: The drone spraying system for snow accumulation recognition and path dynamic planning on grade-separated roads also includes an output and visualization module that interacts with each module. The output and visualization module displays the entire mission process through a graphical interface and generates structured reports that include flight, spraying, and resource consumption. It supports exporting in multiple formats for archiving and analysis.

Images