A coverage path planning method based on area contraction

Abstract

The present application relates to the technical field of UAV path planning, and in particular to a coverage path planning method based on regional shrinkage, which includes an image segmentation module, a boundary feature extraction module, a shrinkage ring generation module, a path optimization module, and a path selection module. This method generates step-by-step shrinkage rings and key nodes by performing semantic segmentation and boundary feature extraction on high-resolution images, and optimizes the path by path smoothing and redundant node elimination, and finally dynamically selects the optimal coverage path according to the task requirements. The present application does not require pre-decomposition of non-convex or mixed areas, significantly improves efficiency, combines dynamic weight allocation with real-time adjustment mechanism, supports multi-objective optimization path selection, reduces redundant flights and coverage blind spots, and uses cubic spline interpolation algorithm to improve flight stability. It is suitable for scenes such as agricultural spraying and terrain mapping, and has strong practicality and promotion value.

Description

Technical Field

[0001] The present invention belongs to the technical field of path planning, and in particular relates to a coverage path planning method based on area contraction. Background Art

[0002] Coverage path planning technology is widely used in fields such as drone inspections, agricultural plant protection, and environmental monitoring. Its core goal is to achieve efficient, full coverage of a target area through optimized path design. However, traditional coverage path planning algorithms have significant shortcomings when dealing with irregular regions, such as non-convex or multiply connected regions. These methods typically rely on manual annotation to identify region boundaries and fail to fully utilize multi-scale feature extraction techniques from high-resolution imagery. This results in inefficient boundary identification and difficulty in dynamically updating coverage parameters.

[0003] Chinese patent publication number CN116797665A discloses a drone adaptive coverage path planning method for aircraft surface inspection, which uses a drone-mounted camera to perform visual inspection of aircraft surface damage; the viewpoint generation module performs surface adaptive discretization sampling based on an existing aircraft point cloud model to generate a pose sample set and a corresponding viewpoint sample set, and screens the samples based on distance, collision, and coverage area; the search space construction module connects the screened pose samples to construct a drone search space; the path search module performs path search in the search space based on the constructed weighted heuristic reward function; the coverage evaluation module uses the voxel volume method to calculate the coverage rate during the path search, and finally generates a drone damage inspection path that meets the target coverage rate.

[0004] In addition, most existing technologies do not combine real-time image data collected by drone cameras (such as terrain features, obstacle distribution, etc.) to dynamically adjust path parameters, which can easily lead to redundant flights or coverage blind spots, seriously affecting coverage efficiency and operation quality.

[0005] On the other hand, for coverage tasks in non-convex areas or mixed convex and non-convex areas, traditional methods usually require complex regional pre-decomposition operations, dividing the target area into multiple regular sub-areas and then planning paths for each. This pre-decomposition process not only increases the computational complexity, but may also affect the coverage effect due to inaccurate decomposition. At the same time, existing methods lack the utilization of the step-by-step contraction characteristics of the region when generating coverage paths, making it difficult to form an efficient coverage path by connecting the key nodes on the contraction rings at each level. Therefore, developing a coverage path planning method that does not require regional pre-decomposition and can adaptively handle convex and non-convex areas has become a technical problem that needs to be solved urgently.

[0006] This invention aims to address these issues by integrating image segmentation technology with a path planning algorithm. This method utilizes real-time image data collected by drones to dynamically generate region boundary features. This method, combined with a step-by-step shrinking ring generation method, optimizes the calculation of virtual offset vertices, thereby improving the efficiency and accuracy of coverage path planning. This method not only avoids the need for pre-decomposition of irregular regions but also dynamically adjusts path parameters based on real-time image information, effectively reducing redundant flights and coverage blind spots. This provides a more efficient and reliable solution for coverage tasks in complex scenarios. Summary of the Invention

[0007] The purpose of the present invention is to provide a coverage path planning method based on area contraction to solve the problems mentioned in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A coverage path planning method based on region shrinkage is characterized by comprising an image segmentation module, a boundary feature extraction module, a shrinkage ring generation module, a path optimization module and a path selection module; the image segmentation module is used to perform semantic segmentation on high-resolution images collected by a drone; the boundary feature extraction module is used to extract region boundary features from the segmented image; the shrinkage ring generation module is used to generate step-by-step shrinkage rings and calculate the positions of key nodes; the path optimization module is used to smooth the generated path and eliminate redundant nodes; and the path selection module is used to select the optimal coverage path according to task requirements.

[0010] As a further solution of the present invention, the shrinkage ring generation module includes a virtual offset vertex calculation unit; the virtual offset vertex calculation unit is used to dynamically adjust the offset distance according to the boundary characteristics and generate the key node coordinates of the shrinkage ring step by step. The formula is as follows:

[0011] ;

[0012] in, Indicates the The first stage shrink ring Coordinates of the nodes (front contraction ring nodes), is the offset step size, represents the coordinates of the i-th node on the k-1-th level contraction ring (the node of the previous level contraction ring),

[0013] Indicates that the boundary characteristic function is at the node The gradient direction at .

[0014] As a further solution of the present invention, the shrink ring generation module further includes a boundary constraint correction unit; the boundary constraint correction unit is used to perform boundary integrity verification on the generated shrink ring and correct the node positions that exceed the boundary. The formula is as follows:

[0015] ;

[0016] in, Representation node To the border distance, is the maximum allowable deviation value, is the correction factor, are the corrected node coordinates.

[0017] As a further solution of the present invention, the path optimization module includes a redundant node elimination unit; the redundant node elimination unit is used to eliminate redundant nodes based on the Euclidean distance and angle relationship between adjacent nodes, and the formula is as follows:

[0018] ;

[0019] when and When , remove the node ,in For nodes The angle between is the reference node (the starting point for comparison), is the node to be evaluated (redundant point that may be eliminated), For subsequent adjacent nodes ( The formula calculates the steering angle formed by three consecutive path nodes. is the minimum angle threshold, , for nodes The Euclidean distance of is the maximum distance threshold.

[0020] As a further solution of the present invention, the path optimization module further includes a path smoothing unit; the path smoothing unit is used to smooth the optimized path and generate a smooth path using a cubic spline interpolation algorithm, and the formula is as follows:

[0021] ;

[0022] in, , , , is the interpolation coefficient, which is determined by the node coordinates and path continuity conditions.

[0023] As a further solution of the present invention, the path selection module includes a dynamic weight allocation unit; the dynamic weight allocation unit is used to ;

[0024] in, represents the path length cost, Indicates the time cost, represents the energy cost, , , , is the weight coefficient.

[0025] As a further solution of the present invention, the path selection module further includes a real-time adjustment unit; the real-time adjustment unit is used to update the path weight in real time according to the drone sensor data

[0026] As a further solution of the present invention, the path smoothing unit further adopts an anti-interference cubic spline interpolation algorithm, and the specific calculation formula is:

[0027] ;

[0028] in, is the anti-interference coefficient, is the noise suppression term, is the noise point weight, dynamically estimated by the Kalman filter, The highest order of the main path polynomial determines the complexity of the basic spline. is the number of noise points detected, is the spline interpolation coefficient, Parameterize the path variable. is the parameter position of the noise point.

[0029] As a further solution of the present invention, the dynamic weight allocation unit further introduces an energy recovery cost item , the specific calculation formula is:

[0030] ;

[0031] in, The UAV dynamics model is used to calculate the kinetic energy recovery during descent or turning, which satisfies , is the energy conversion efficiency, For drone quality, is the instantaneous speed.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The present invention does not require pre-decomposition of non-convex or mixed-type areas, and directly generates coverage paths through step-by-step shrinkage rings, significantly improving the efficiency of the algorithm. Secondly, it combines image segmentation technology to dynamically extract boundary features, avoiding reliance on manual annotation and improving the degree of automation of path planning. Thirdly, through dynamic weight allocation and real-time adjustment mechanisms, it supports multi-objective optimization path selection in complex environments, reducing redundant flights and coverage blind spots. In addition, the use of a cubic spline interpolation algorithm to generate a smooth path improves the flight stability of the drone and reduces mechanical wear. This method is applicable to a variety of scenarios, such as agricultural spraying, terrain mapping, etc., and has strong practicality and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 : Schematic diagram of the process of the present invention;

[0035] Figure 2 : Schematic diagram of the contractile ring generation module of the present invention;

[0036] Figure 3 : Processing flow chart of the path optimization module of the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0038] Reference Figure 1-Figure 3 As shown in the figure, the present invention provides a coverage path planning method based on area shrinkage, which is suitable for path planning tasks of UAVs in complex scenarios, such as agricultural spraying, terrain mapping and other fields. First, starting from the overall process, the specific implementation methods of the image segmentation module, boundary feature extraction module, shrinkage ring generation module, path optimization module and path selection module are explained, and in-depth analysis is conducted in combination with actual application scenarios.

[0039] In the specific implementation process, the operation process of the entire method is as follows Figure 1As shown, it includes five main functional modules: image segmentation module, boundary feature extraction module, contraction ring generation module, path optimization module and path selection module. First, the UAV collects image data of the target area through the high-resolution camera it carries. These image data are transmitted to the image segmentation module, which uses semantic segmentation technology to process the image and divide the target area into different semantic categories, such as farmland area, obstacle area, etc. In order to improve the segmentation accuracy, the deep learning model U-Net is used as the core algorithm in this embodiment, which has strong feature extraction capabilities and pixel-level segmentation effects. The segmented image provides basic data support for subsequent boundary feature extraction.

[0040] The boundary feature extraction module is responsible for extracting regional boundary features from the segmented image. The extraction of boundary features is completed by calculating the image gradient and combining it with the edge detection algorithm. Specifically, in this embodiment, the Sobel operator is used to calculate the gradient of the segmented image to obtain the gradient direction and gradient amplitude of each pixel. Subsequently, the non-maximum suppression and double threshold detection methods are used to further screen out significant boundary points. These boundary points constitute the contour information of the target area and lay the foundation for the subsequent step-by-step contraction ring generation. The output result of the boundary feature extraction module is a set containing the coordinates of the boundary points, which will be passed as input to the contraction ring generation module.

[0041] The core task of the shrinking ring generation module is to generate shrinking rings step by step based on boundary features and calculate the locations of key nodes on each shrinking ring. This module includes a virtual offset vertex calculation unit and a boundary constraint correction unit. The virtual offset vertex calculation unit dynamically adjusts the offset distance to ensure that the shrinking ring can adapt to the complex boundaries of non-convex areas. The specific calculation formula is as follows:

[0042] ;

[0043] in, Indicates the The first stage shrink ring The coordinates of the node, is the offset step size, represents the coordinates of the i-th node on the k-1-th level contraction ring (the node of the previous level contraction ring), Indicates that the boundary characteristic function is at the node The formula dynamically adjusts the offset step size by the gradient direction to ensure that the contraction ring can closely fit the boundary of the target area. In practical applications, the offset step size The value of needs to be adjusted according to the size and complexity of the target area, and is usually set between 0.5 and 2 meters. In addition, the boundary constraint correction unit is used to check the boundary integrity of the generated shrinkage ring and correct the position of nodes that exceed the boundary. The correction formula is as follows:

[0044] ;

[0045] in, Representation node To the border distance, is the maximum allowable deviation value, is the correction factor, The above formula can effectively avoid the situation where the nodes of the shrinkage ring deviate from the actual boundary, thereby ensuring that the generated shrinkage ring is consistent with the boundary of the target area.

[0046] The main task of the path optimization module is to smooth the generated path and remove redundant nodes to improve the feasibility and efficiency of the path. This module includes a redundant node removal unit and a path smoothing unit. The redundant node removal unit removes unnecessary redundant nodes by calculating the Euclidean distance and angle relationship between adjacent nodes. The specific calculation formula is as follows:

[0047] ;

[0048] when and When , remove the node ,in For nodes The angle between is the reference node (the starting point for comparison), is the node to be evaluated (redundant point that may be eliminated), For subsequent adjacent nodes ( The formula calculates the steering angle formed by three consecutive path nodes. is the minimum angle threshold, , for nodes The Euclidean distance of is the maximum distance threshold. This formula effectively reduces path redundancy through the dual constraints of angle and distance. In practical applications, the angle threshold and distance threshold The value of needs to be adjusted according to the flight performance and mission requirements of the UAV. For example, in the agricultural spraying scenario, Can be set to 30 degrees, It can be set to 5 meters. The path smoothing unit uses the cubic spline interpolation algorithm to smooth the optimized path and generate a smooth path. The interpolation formula is as follows:

[0049] ;

[0050] in, , , , is the interpolation coefficient, which is determined by the node coordinates and the path continuity condition. This algorithm ensures that the path has second-order continuity between nodes, improving the flight stability of the drone.

[0051] Furthermore, during drone flight, path smoothing is susceptible to sensor noise or environmental interference (such as electromagnetic interference and false obstacle detection), resulting in path jitter. The anti-interference cubic spline interpolation algorithm is used to suppress noise and improve path robustness. The path smoothing unit further uses the anti-interference cubic spline interpolation algorithm. The specific calculation formula is:

[0052] ;

[0053] in, is the anti-interference coefficient, with a value range of 0.1-0.5 (determined through experiments, a larger value is taken in a high noise environment), The highest order of the main path polynomial determines the complexity of the basic spline. is the number of noise points detected, is the spline interpolation coefficient, Parameterize the path variable. is the parameter position of the noise point, The noise suppression term is the Dirac function, which means that at the noise point Apply suppression, is the noise point weight, which is dynamically estimated by the Kalman filter. The initial value is set to 1 and is adaptively adjusted with the noise intensity. The Kalman filter dynamically estimates the position deviation of the input noise point. (the difference between the current coordinates and the historical coordinates), through the state equation:

[0054] ;

[0055] in, is the Kalman gain filter gain, determined by the noise covariance matrix, is the observation matrix, thus outputting the updated weight , is the noise point position deviation, if <0.2, it is determined to be an invalid noise point and is not suppressed. The interpolated path The data is sent to the UAV flight control system to correct the flight trajectory in real time. For example, in a farmland spraying scenario, the UAV's path may be jittered due to wind interference. This algorithm can reduce the maximum path offset by 60%. Through noise detection, anti-interference interpolation and dynamic weight estimation of Kalman filtering, the path jitter is significantly reduced, making it suitable for high-interference environments.

[0056] The path selection module is responsible for selecting the optimal coverage path based on mission requirements and adjusting path weights in real time to cope with dynamic environmental changes. This module includes a dynamic weight allocation unit and a real-time adjustment unit. The dynamic weight allocation unit calculates the total path weight using the following formula:

[0057] ;

[0058] in, represents the path length cost, Indicates the time cost, represents the energy cost, , , , is the weight coefficient. This formula realizes multi-objective optimization path selection through dynamic weight allocation. In practical applications, the value of the weight coefficient needs to be adjusted according to the task priority. For example, in an emergency rescue scenario, the time cost Weight should be set to a higher value; in energy-constrained scenarios, the energy cost Weight It should be set to a high value. The real-time adjustment unit updates the path weight in real time based on the drone sensor data to ensure the robustness of path planning in dynamic environments.

[0059] When the drone is descending or turning, part of the kinetic energy can be recovered into electrical energy through motor reverse drag or potential energy conversion, which can be achieved by introducing energy recovery cost items. , optimize path selection to extend endurance Dynamic weight allocation unit further introduces energy recovery cost item , the specific calculation formula is:

[0060] ;

[0061] in, The UAV dynamics model is used to calculate the kinetic energy recovery during descent or turning, which satisfies , is the energy conversion efficiency, For drone quality, is the instantaneous speed (obtained in real time by IMU). The specific weight coefficient is dynamically adjusted according to the mission mode, such as emergency mode: =0.7, =0.2, =0.1 (prioritize time cost); energy saving mode: =0.6, =0.3, =0.1 (prioritize energy consumption optimization); when the UAV is mapping in the mountains, it chooses a path that includes a descending slope (recoverable energy),

[0062] Calculate the total weight of the two candidate paths:

[0063] Path A (steep descent): ;

[0064] Path B (level flight): .

[0065] like =0.6, path A has a lower weight and is preferred. The battery management system (BMS) monitors the actual recovered energy in real time, compares it with the model prediction value, and dynamically corrects the μ value. By combining the energy recovery model with dynamic weight allocation, the path is optimized to extend the battery life. This is suitable for long-term inspections or energy-sensitive scenarios.

[0066] Through the collaborative work of the above five modules, the present invention realizes the complete process from image segmentation to optimal path selection. The specific implementation steps are as follows: First, the UAV collects high-resolution images of the target area and performs semantic segmentation through the image segmentation module; then, the boundary feature extraction module extracts the regional boundary features and passes them to the contraction ring generation module; then, the contraction ring generation module generates step-by-step contraction rings and calculates the positions of key nodes; then, the path optimization module smoothes the generated path and eliminates redundant nodes; finally, the path selection module selects the optimal coverage path according to the task requirements, and dynamically updates the path weight through the real-time adjustment unit. The whole process is as follows Figure 1 As shown in the figure, the modules are closely connected, ensuring the efficiency and accuracy of path planning.

[0067] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A coverage path planning method based on area contraction, characterized in that: It includes an image segmentation module, a boundary feature extraction module, a contraction ring generation module, a path optimization module and a path selection module; the image segmentation module is used to perform semantic segmentation on the high-resolution image collected by the drone; the boundary feature extraction module is used to extract regional boundary features from the segmented image; The shrinkage ring generation module is used to generate step-by-step shrinkage rings and calculate the positions of key nodes; the path optimization module is used to smooth the generated path and eliminate redundant nodes; The path selection module is used to select the optimal coverage path according to task requirements; The shrinkage ring generation module includes a virtual offset vertex calculation unit; the virtual offset vertex calculation unit is used to dynamically adjust the offset distance according to the boundary characteristics and generate the key node coordinates of the shrinkage ring step by step. The formula is as follows: ； in, Indicates the The first stage shrink ring The coordinates of the node, is the offset step size, represents the coordinates of the i-th node on the k-1-th level contractile ring, Indicates that the boundary characteristic function is at the node The gradient direction at ; The shrink ring generation module further includes a boundary constraint correction unit; the boundary constraint correction unit is used to perform boundary integrity verification on the generated shrink ring and correct the node positions that exceed the boundary. The formula is as follows: ； in, Representation node To the border distance, is the maximum allowable deviation value, is the correction factor, are the corrected node coordinates.

2. The coverage path planning method based on area contraction according to claim 1, characterized in that: The path optimization module includes a redundant node elimination unit; the redundant node elimination unit is used to eliminate redundant nodes based on the Euclidean distance and angle relationship between adjacent nodes. The formula is as follows: ； when and When , remove the node ,in For nodes The angle between is the reference node, is the node to be evaluated, is the subsequent adjacent node, is the minimum angle threshold, For nodes The Euclidean distance of is the maximum distance threshold.

3. The coverage path planning method based on area contraction according to claim 2, characterized in that: The path optimization module also includes a path smoothing unit; the path smoothing unit is used to smooth the optimized path and generate a smooth path using a cubic spline interpolation algorithm. The formula is as follows: ； in, , , , is the interpolation coefficient, which is determined by the node coordinates and path continuity conditions, Parameterize variables for paths.

4. The coverage path planning method based on area contraction according to claim 1, characterized in that: The path selection module includes a dynamic weight allocation unit; the dynamic weight allocation unit is used to ; in, represents the path length cost, Indicates the time cost, represents the energy cost, , , is the weight coefficient.

5. The coverage path planning method based on area contraction according to claim 4, characterized in that: The path selection module also includes a real-time adjustment unit; the real-time adjustment unit is used to update the path weight in real time according to the drone sensor data.

6. The coverage path planning method based on area contraction according to claim 3, characterized in that: The path smoothing unit further adopts an anti-interference cubic spline interpolation algorithm, and the specific calculation formula is: ； in, is the anti-interference coefficient, is the noise suppression term, is the highest order of the main path polynomial, is the number of noise points detected, is the spline interpolation coefficient, is the parameter position of the noise point, indicating that Apply suppression, is the noise point weight.

7. The coverage path planning method based on area contraction according to claim 5, characterized in that: The dynamic weight allocation unit further introduces an energy recovery cost term , the specific calculation formula is: ； in, The UAV dynamics model is used to calculate the kinetic energy recovery during descent or turning, which satisfies , is the energy conversion efficiency, For drone quality, is the instantaneous speed.

Images