Epidemic prevention robot path planning method fused with Voronoi skeleton diagram RRT algorithm

Through the RRT algorithm that integrates the Voronoi skeleton diagram, the problem of low path planning efficiency in complex environments is solved, efficient and stable path planning is achieved, and path quality and time consumption are optimized.

CN120255511APending Publication Date: 2025-07-04GUILIN UNIV OF ELECTRONIC TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510379652.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-03-28
Publication Date
2025-07-04

AI Technical Summary

Technical Problem

The existing RRT algorithms have low path planning efficiency in complex environments, making it difficult to obtain high-quality paths in a short time, and there are redundant exploration problems.

Method used

Using the RRT algorithm that fuses Voronoi skeleton graph, the initial path is obtained by constructing the Voronoi skeleton graph, subdividing it into subpaths and guiding the expansion, combining elliptical constraints and gravitational field biasing strategies, and using a multi-segment pruning strategy to smooth the path.

Benefits of technology

It improves the efficiency of path planning, shortens the path length and number of inflection points, reduces the number of samples, optimizes the planning time, and maintains the stability and environmental adaptability of the algorithm.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120255511A_ABST
    Figure CN120255511A_ABST
Patent Text Reader

Abstract

The invention discloses an epidemic prevention robot path planning method fused with a Voronoi skeleton diagram RRT algorithm. The method comprises the following specific implementation steps: step 1, acquiring and connecting equidistant points among map obstacles to construct a Voronoi skeleton diagram, and storing the Voronoi skeleton diagram locally; 2, acquiring an initial heuristic path on the skeleton diagram by adopting an A * algorithm; step 3, based on the initial path node, subdividing the global path into a plurality of sub-paths, and guiding the RRT algorithm to extend to sub-target nodes one by one; step 4, constructing an elliptical constraint channel for each sub-path to perform non-uniform sampling; 5, redundant exploration of an RRT algorithm is avoided in combination with a gravitational field bias strategy; and 6, after the planning is completed, sequentially connecting each sub-path as a result path, and smoothing a final path by using a multi-segment pruning strategy. Compared with an existing algorithm, the method has the advantages of path smoothness, high operation efficiency and environmental adaptability, and the epidemic prevention robot can stably and quickly execute a planning task.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of path planning of intelligent robots, and particularly relates to a path planning method for epidemic prevention robots that integrates the Voronoi skeleton graph RRT algorithm. Background Technique

[0002] In recent years, the technology of epidemic prevention robots in special robots has become increasingly mature in China and has been widely applied in various traditional industries and enterprises. At present, epidemic prevention robots are mainly applied in occasions such as medical institutions, event venues, urban public areas, and factories. The epidemic prevention robots in these occasions have their own control characteristics, such as low driving speed, complex driving trajectory, large curvature change, simple kinematic characteristics, etc. In some cases, epidemic prevention robots are also affected by various unstructured roads, non-traffic rule constraints, and non-artificially controlled factors. Therefore, how to improve the autonomous navigation ability of epidemic prevention robots in these complex environments is a problem that needs to be studied currently.

[0003] Due to its random sampling characteristics, the traditional RRT algorithm is prone to the problem of robustness caused by relying on a large number of samples for the algorithm to explore effective areas when dealing with complex environments; in the increasingly extensive working scenarios of epidemic prevention robots, the RRT algorithm faces a trade-off between real-time performance and path quality: the better the path quality, the longer the time consumption often means. Obtaining a higher-quality path in a shorter time is the requirement that the RRT algorithm should meet. Therefore, solving the problems of difficult growth and low convergence efficiency faced by the RRT algorithm in complex environments and helping the RRT algorithm to achieve efficient and stable path planning in complex scenarios is of great significance for the application and development of epidemic prevention robots. Summary of the Invention

[0004] The following is an overview of the subject matter described in detail in this article. This overview is not intended to limit the scope of protection of the claims.

[0005] In view of the deficiencies of the prior art, the present invention proposes a path planning method for epidemic prevention robots that integrates the Voronoi skeleton graph RRT algorithm, which improves the path planning efficiency and environmental adaptability.

[0006] According to a path planning method for epidemic prevention robots that integrates the Voronoi skeleton graph RRT algorithm of the present invention, it includes the following steps:

[0007] Step 1: Obtain and connect equidistant points between map obstacles to construct a Voronoi skeleton graph and store it locally;

[0008] Step 2: Use the A* algorithm to obtain an initial heuristic path on the skeleton graph;

[0009] Step 3: Based on the initial path nodes, divide the global path into multiple sub-paths and guide the RRT algorithm to expand towards the sub-goal nodes;

[0010] Step 4: Construct an elliptical constraint channel for each sub-path for non-uniform sampling;

[0011] Step 5: Combine the gravitational field biasing strategy to avoid redundant exploration of the RRT algorithm;

[0012] Step 6: After the planning is completed, connect each sub-path in sequence as the result path and use a multi-segment pruning strategy to smooth the final path.

[0013] Preferably, in Step 1, a construction method based on scan lines is adopted to sequentially obtain and connect the set of equidistant points between obstacles from left to right, and dynamically construct the Voronoi diagram boundary. And store the vertex coordinates of the Voronoi diagram and their adjacency relationships offline in the form of an adjacency matrix. For the map settings, three maps with different scenarios are set for testing, namely: dense obstacle scenario, maze scenario, and office scenario.

[0014] Preferably, in Step 2, determine the two Voronoi vertices closest to the given start point and end point, and use them as the start point and end point of the A* algorithm respectively. Use the A* algorithm to quickly find a shortest initial path in the Voronoi skeleton diagram.

[0015] Preferably, in Step 3, the initial path consists of a set of Voronoi vertices, which divide the entire path into multiple sub-paths, guide the random tree to expand in each sub-path, and there is no data dependency between the path planning problems within each sub-path, and multiple threads are used to process.

[0016] Preferably, in Step 4, an elliptical constraint is used to construct the safety region of each sub-path. By defining the start point and end point of the sub-path as the foci of the elliptical region, the sampling region is concentrated between the start point and the end point. And considering the probabilistic completeness of the RRT algorithm, a bias value α is introduced. When the random number is greater than α, the random tree will perform random sampling outside the elliptical region. When the generated random number is less than α, the algorithm randomly generates a sample point p(X rand , Y rand , 1) in the standard coordinate system of the ellipse and converts it into a point inside the ellipse in the current map coordinate system by multiplying by the transformation matrix T.

[0017] Preferably, in Step 5, the gravitational field biasing idea of the artificial potential field is adopted to make the algorithm further converge. During the process of randomly generating new nodes, by applying an attractive force to the target position of the current sub-path, the newly generated nodes are more inclined to move towards the target node. The newly generated node x newFinally, it will grow in the direction of the combined action of the random component and the gravitational potential field.

[0018] In step six, an adaptive multi-segment pruning strategy is adopted. The end point of the path is used as the pruning starting point, and the parent node of the path end point is tentatively set as the pruning end point. Determine whether the current pruning starting point and end point are collision-free. If there is no collision, move the pruning end point to the next node. By traversing the nodes in the path until a node that satisfies no collision and is the farthest is found, it is used as the real pruning end point for pruning. During the pruning process, the original nodes and their connection relationships between the starting point and the end point are retained, and they are moved to the line connecting the starting point and the end point divided equally by the distance. After pruning is completed, continue the next pruning operation. Move the pruning starting point to its parent node, and the pruning end point repeats the above operation until a new node that meets the requirements is found to perform a new pruning operation. And so on, the pruning end point will eventually move to the starting point of the whole path, and perform the last pruning operation on it, then the pruning is completed. Finally, a smooth optimal path without redundant inflection points can be obtained.

[0019] The beneficial effects of the present invention are as follows: The present invention starts from the map construction layer, extracts key skeleton information from the Voronoi diagram to initialize the given environment, and quickly obtains an initial heuristic path based on the skeleton information. This heuristic path is used to guide the sampling process of the RRT. On this basis, path points are selected to divide the initial path into multiple sub-paths, an elliptical constraint is introduced to limit the sampling range, and a gravitational field biasing strategy is combined to avoid redundant exploration and shorten the planning time. An adaptive multi-segment pruning strategy is proposed to smooth the path. Simulation experiments show that the improved RRT path planning algorithm integrating the Voronoi skeleton diagram of the present invention realizes a 19.53%, 20.55%, and 26.2% reduction in path length, a 92.49%, 68.25%, and 92.72% reduction in path inflection points, and a 24.49%, 46.03%, and 17.43% reduction in the number of samples in three scenarios respectively. At the same time, the optimization amplitude of the planning time is maintained at more than 41% in scenarios two and three. Compared with the existing methods, the present invention maintains the probabilistic completeness of the sampling-based algorithms and obtains stable performance close to that of the graph search algorithms. Brief Description of the Drawings

[0020] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solutions of the present invention and do not constitute a limitation to the technical solutions of the present invention.

[0021] Figure 1 It is a flowchart of the path planning method for the epidemic prevention robot of the RRT algorithm integrating the Voronoi skeleton diagram of the present invention;

[0022] Figure 2It is a schematic diagram of the map topology structure and the initial path of the path planning method for the epidemic prevention robot that integrates the Voronoi skeleton graph RRT algorithm of the present invention;

[0023] Figure 3 It is a schematic diagram of elliptical constraint sampling of the path planning method for the epidemic prevention robot that integrates the Voronoi skeleton graph RRT algorithm of the present invention;

[0024] Figure 4 It is a schematic diagram of the gravitational field structure of the path planning method for the epidemic prevention robot that integrates the Voronoi skeleton graph RRT algorithm of the present invention;

[0025] Figure 5 It is a schematic diagram of the pruning strategy principle of the path planning method for the epidemic prevention robot that integrates the Voronoi skeleton graph RRT algorithm of the present invention;

[0026] Figure 6 It is a graph of the final optimized path planning result in an obstacle-dense scenario provided by an embodiment of the present invention;

[0027] Figure 7 It is a graph of the final optimized path planning result in a maze scenario provided by an embodiment of the present invention;

[0028] Figure 8 It is a graph of the final optimized path planning result in an office scenario provided by an embodiment of the present invention. Detailed implementation manners

[0029] To further understand the content of the present invention, the present invention will be described in detail in combination with the accompanying drawings and embodiments. It should be understood that the embodiments are only for explaining the present invention rather than limiting it.

[0030] According to a path planning method for an epidemic prevention robot that integrates the Voronoi skeleton graph RRT algorithm of the present invention, it includes the following steps:

[0031] Step 1: Use Python to establish three test maps with different complexities, obtain and connect the equidistant points between the obstacles in the map to construct a Voronoi skeleton graph, extract the key vertices and their adjacency relationships, and store them offline using an adjacency matrix;

[0032] Step 2: Determine the two Voronoi vertices closest to the given starting point and ending point, and use them as the starting point and ending point of the A* algorithm respectively to quickly find a shortest initial path in the Voronoi skeleton graph;

[0033] Step 3: Based on the initial path nodes, divide the global path into multiple sub-paths, use multiple threads to process the guidance, and the RRT algorithm expands towards the sub-goal nodes;

[0034] Step 4: Define the starting and ending points of the sub-path as the foci of the ellipse to construct the safety region of each sub-path. A bias value α is introduced. When the random number is greater than α, the random tree will perform random sampling outside the elliptical region. When the generated random number is less than α, the algorithm randomly generates a sample point p(X rand , Y rand , 1) in the standard coordinate system of the ellipse, and converts it into a point inside the ellipse in the current map coordinate system by multiplying with the transformation matrix T. The overall sampling process of the algorithm can be described as:

[0035]

[0036] where p represents a probability uniformly sampled from 0 to 1, and Sample() represents sampling irregularly outside the ellipse.

[0037] The sampling process inside the ellipse can be described as:

[0038]

[0039] x ellipse = T × p T

[0040] where a represents the semi-major axis of the ellipse, b represents the semi-minor axis of the ellipse, h represents the X-axis coordinate of the origin of the ellipse, k represents the Y-axis coordinate of the origin of the ellipse, and θ represents the current starting and ending direction of the ellipse.

[0041] Step 5: During the process of randomly generating new nodes, apply an attraction force to the target position of the current sub-path, so that the newly generated nodes tend to move towards the target node. The newly generated node x new will eventually grow in the direction of the combined action of the random component and the gravitational potential field. The specific bias strategy is expressed as:

[0042]

[0043] x new = x nearest + F

[0044] where ε represents the gravitational coefficient, and the larger the value, the stronger the gravitational force. step represents the step size, represents the vector in the random direction, represents the vector in the direction of the gravitational potential field, and F is the magnitude of the final resultant force.

[0045] Step 6: After the planning is completed, connect each sub-path in sequence as the result path, and use a multi-segment pruning strategy to smooth the final path. Take the end point of the path as the pruning starting point, and temporarily set the parent node of the path end point as the pruning end point. Determine whether the current pruning starting point and end point are connected without collision. If there is no collision, move the pruning end point to the next node. By traversing the nodes in the path until a node that satisfies no collision and is the farthest away is found, use it as the real pruning end point for pruning. During the pruning process, retain the original nodes and their connection relationships between the starting point and the end point, and move them to the line equidistant from the starting point and the end point. After the pruning is completed, continue the next pruning operation. Move the pruning starting point to its parent node, and repeat the above operation for the pruning end point until a new node that meets the requirements is found for a new pruning operation. And so on, the pruning end point will finally move to the starting point of the whole path, and perform the last pruning operation on it, then the pruning is completed.

[0046] The above is a specific description of the preferred embodiment of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A path planning method for epidemic prevention robots integrating the Voronoi skeleton graph RRT algorithm, characterized in that Implement path planning by using the Voronoi skeleton graph to guide the RRT algorithm, including the following steps: Step 1: Obtain and connect the equidistant points between the map obstacles to construct the Voronoi skeleton graph and store it locally; Step 2: Use the A* algorithm to obtain the initial heuristic path on the skeleton graph; Step 3: Based on the initial path nodes, divide the global path into multiple sub-paths to guide the RRT algorithm to expand towards the sub-goal nodes; Step 4: Construct an elliptical constraint channel for each sub-path for non-uniform sampling; Step 5: Combine the gravitational field biasing strategy to avoid redundant exploration of the RRT algorithm; Step 6: After the planning is completed, connect each sub-path in sequence as the result path, and use a multi-segment pruning strategy to smooth the final path.

2. The epidemic prevention robot path planning method integrating the Voronoi skeleton graph RRT algorithm according to claim 1, characterized in that: In Step 1, a scan-line-based construction method is adopted. The equidistant point sets between the obstacles are obtained and connected in sequence from left to right to dynamically construct the Voronoi diagram boundary. The vertex coordinates of the Voronoi diagram and their adjacency relationships are stored offline in the form of an adjacency matrix. For the map settings, three maps with different scenarios are set for testing, namely: the obstacle-dense scenario, the maze scenario, and the office scenario.

3. The epidemic prevention robot path planning method integrating the Voronoi skeleton graph RRT algorithm according to claim 1, characterized in that: In Step 2, determine the two Voronoi vertices closest to the given start point and end point, and use them as the start point and end point of the A* algorithm respectively. Use the A* algorithm to quickly find a shortest initial path in the Voronoi skeleton graph.

4. The epidemic prevention robot path planning method integrating the Voronoi skeleton graph RRT algorithm according to claim 1, characterized in that: In Step 3, the initial path consists of a set of Voronoi vertices. These vertices divide the entire path into multiple sub-paths, guiding the random tree to expand in each sub-path. There is no data dependency between the path planning problems within each sub-path, and multiple threads are used to handle it.

5. The epidemic prevention robot path planning method integrating the Voronoi skeleton graph RRT algorithm according to claim 1, characterized in that: In Step 4, an elliptical constraint is used to construct the safety region of each sub-path. By defining the start point and end point of the sub-path as the foci of the elliptical region, the sampling region is concentrated between the start point and the end point. Considering the probabilistic completeness of the RRT algorithm, a bias value α is introduced. When the generated random number is less than α, the random tree will perform random sampling in the elliptical region. When the random number is greater than α, the random tree will perform random sampling outside the elliptical region. The overall sampling process of the algorithm can be described as: Among them, p represents a probability uniformly sampled from 0 to 1, and Sample() represents sampling irregularly outside the ellipse. The sampling process inside the ellipse can be described as: x ellipse = T × p T The algorithm randomly generates a sample point p(X rand , Y rand , 1) in the standard coordinate system of the ellipse, and converts it into an internal point of the ellipse in the current map coordinate system by multiplying by the transformation matrix T, where a represents the semi-major axis of the ellipse, b represents the semi-minor axis of the ellipse, h represents the X-axis coordinate of the origin of the ellipse, k represents the Y-axis coordinate of the origin of the ellipse, and θ represents the given current starting and ending directions of the ellipse.

6. The epidemic prevention robot path planning method integrating the Voronoi skeleton graph RRT algorithm according to claim 1, characterized in that: In Step 5, the gravitational field biasing idea of the artificial potential field is adopted to further converge the algorithm. During the process of randomly generating new nodes, an attractive force is applied to the target position of the current sub-path, making the newly generated nodes more inclined to move towards the target node. The newly generated node x new will eventually grow in the direction of the combined action of the random component and the gravitational potential field. The specific biasing strategy is expressed as: x new = x nearest + F Among them, ε represents the gravitational coefficient, and the larger the value, the stronger the gravitational force. step represents the step size, represents the vector in the random direction, represents the vector in the direction of the gravitational potential field, and F is the magnitude of the final resultant force.

7. The epidemic prevention robot path planning method integrating the Voronoi skeleton graph RRT algorithm according to claim 1, characterized in that: In step six, an adaptive multi-segment pruning strategy is adopted. Taking the end point of the path as the pruning starting point, the parent node of the path end point is tentatively set as the pruning end point. Determine whether the current pruning starting point and end point are connected without collision. If there is no collision, move the pruning end point to the next node. By traversing the nodes in the path until a node that satisfies no collision and is the farthest is found, and use it as the real pruning end point for pruning. During the pruning process, retain the original nodes and their connection relationships between the starting point and the end point, and move them to the line equidistant from the starting point and the end point. After pruning, continue the next pruning operation. Move the pruning starting point to its parent node, and the pruning end point repeats the above operation until a new node that meets the requirements is found for a new pruning operation. And so on, the pruning end point will finally move to the starting point of the whole path, and perform the last pruning operation on it, then the pruning is completed. Finally, a smooth optimal path without redundant inflection points can be obtained.