Optimal path planning algorithm of power robot

Abstract

The invention discloses an optimal path planning algorithm for a power robot, and the algorithm comprises the following steps: environment modeling, path planning algorithm, experiment and verification, and the environment modeling comprises the steps: collecting data through a sensor, processing a point cloud and an image, and generating a three-dimensional map for identifying a power transmission line and an obstacle; the path planning algorithm comprises the steps of combining an A star algorithm and a reinforcement learning method, firstly performing initial path planning through the A star algorithm, and then performing dynamic optimization and adjustment on the path through the reinforcement learning method; the method has the advantages that by combining the A star algorithm and the reinforcement learning method, the problems that a traditional path planning algorithm is low in efficiency, poor in dynamic adaptability, lack of real-time obstacle avoidance capacity and the like in a complex environment are solved, efficient and intelligent path planning and dynamic adjustment are achieved, and the reliability and practicability of the algorithm are further guaranteed through comprehensive experimental verification.

Description

Technical Field

[0001] The present invention relates to the technical field of electric robot, and in particular to an optimal path planning algorithm for an electric robot. Background Art

[0002] With the rapid development of society and the continuous increase in electricity demand, the maintenance and inspection of power systems have become increasingly important. As an emerging intelligent device, power inspection robots play an important role in the maintenance and inspection of power systems. Power inspection robots are mainly used to replace manual inspection and maintenance tasks in dangerous and harsh environments to improve inspection efficiency and reduce labor costs and safety risks. However, in a complex power environment, the path planning and real-time obstacle avoidance of power inspection robots have become technical difficulties and research hotspots.

[0003] Traditional path planning algorithms, such as Dijkstra algorithm and A-star algorithm, have been widely used in the field of robot path planning. The A-star algorithm can find the optimal path in a complex environment by combining the actual path cost and heuristic function, and has high computational efficiency and accuracy. However, the A-star algorithm has certain limitations in dynamic environments and is difficult to respond to environmental changes and new obstacles in real time. Therefore, an optimal path planning algorithm for electric robots is proposed. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] The present invention has been proposed in view of the above-mentioned problems.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: an optimal path planning algorithm for an electric robot, comprising the following steps:

[0007] S1, Environmental Modeling: Collect data through sensors, process point clouds and images, and generate three-dimensional maps that identify transmission lines and obstacles;

[0008] S2, path planning algorithm: Combining the A-star algorithm and reinforcement learning method, the initial path planning is first performed through the A-star algorithm, and then the path is dynamically optimized and adjusted through the reinforcement learning method;

[0009] S3, Experiment and Verification: A large number of experiments are carried out in virtual environments and actual power inspection environments to verify the performance of the algorithm.

[0010] As a preferred solution of an optimal path planning algorithm for an electric robot of the present invention, S1 specifically includes:

[0011] S11, uses sensors to collect point cloud data in a temporal environment, the data includes the robot's position, obstacle positions, and other environmental features;

[0012] S12, acquiring point cloud data from the sensor, and converting the point cloud data into a structured three-dimensional map using point cloud processing technology;

[0013] S13, determining the three-dimensional coordinates and geometric shape of the transmission line through measurement and data analysis, and marking it as a specific type of object;

[0014] S14, determining the location and size of the obstacle based on the sensor data and image analysis, and marking it as a three-dimensional shape on the three-dimensional map;

[0015] S15, dividing the three-dimensional map into grid units, each grid unit contains corresponding environmental information, setting a cost value for each grid unit according to the environmental information of each grid unit, and saving the processed three-dimensional map as an STL file.

[0016] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, the environmental information in S15 includes the spatial coordinates of the grid unit, whether there are obstacles in the grid unit, and the terrain in the grid unit.

[0017] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, the initial path planning by the A-star algorithm in S2 specifically includes the following steps:

[0018] S21-1, define the starting grid unit P s and the end grid cell P g , add the starting point to the open list;

[0019] S21-2, take the grid unit with the smallest comprehensive cost from the open list as the current node, record the adjacent grid units of the current node as adjacent nodes, add the current node to the closed list, and for each adjacent node of the current node: if the adjacent node is impassable or is already in the closed list, skip it; if the adjacent node is not in the open list, add it to the open list, calculate its total cost value, and set the current node as its parent node; if the adjacent node is already in the open list, but the new total cost value is smaller, update its total cost value, and set the current node as its parent node, and repeat the above process until the end is found or the open list is empty;

[0020] S21-3, from the end point P g Backtrack to its parent node until the starting point P s, and obtain the path with the minimum cost as the initial path.

[0021] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, the total cost value calculation formula is:

[0022] f(n)=g(n)+h(n)

[0023] Among them, g(n) is the actual cost from the starting point to the current node, which is calculated by accumulating the cost value of each grid cell on the path, and h(n) is the estimated cost from the current node to the end point.

[0024] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, the specific calculation formula of h(n) is:

[0025]

[0026] Among them, (x n ,y n ,z n ) is the current node coordinate, (x g ,y g ,z g ) are the end point coordinates.

[0027] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, the dynamic optimization and adjustment of the path by the reinforcement learning method in S2 specifically includes the following steps:

[0028] S22-1, define an information vector containing information such as the current position of the robot, the position of obstacles, and the target position, define the operations that the robot can perform, including forward, backward, and turning actions, and define a reward function to give corresponding rewards based on whether the robot reaches the final target point, whether it successfully avoids new obstacles and returns to the initial path, whether it collides with obstacles, and the length of the path to avoid obstacles;

[0029] S22-2, establish a deep Q network, including a main network and a target network, the main network is used to select actions, and the target network is used to calculate the target Q value;

[0030] S22-3, training and adjusting main network parameters;

[0031] S22-4, validation in virtual and real environments.

[0032] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, the reward function is specifically:

[0033] R(s,a)=R g +R r +Ro +R l

[0034] Among them, R g is the reward for the robot to reach the final target point. g =100, otherwise R g =0, R r The reward for the robot to avoid the new obstacle and return to the initial path, R o is the penalty for the robot colliding with an obstacle. o = -100, otherwise R o =0, R l The robot is penalized for the length of the path it takes to avoid new obstacles. The longer the path, the greater the penalty value.

[0035] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, when the robot starts to avoid new obstacles and its position coordinates coincide with the initial path at a certain moment, R r =50, otherwise R r =0.

[0036] As a preferred solution of the optimal path planning algorithm of an electric robot of the present invention, R l The calculation formula is:

[0037] R l = -α·d(P t1 ,P t2 )

[0038] Among them, α is the penalty coefficient of path length, d(P t1 ,P t2 ) is the distance the robot moves from the beginning of avoiding the new obstacle to returning to the initial path.

[0039] The beneficial effects of the present invention are as follows: by combining the A-star algorithm and the reinforcement learning method, not only the problems of low efficiency, poor dynamic adaptability, and lack of real-time obstacle avoidance capabilities of traditional path planning algorithms in complex environments are solved, but also efficient and intelligent path planning and dynamic adjustment are realized. Comprehensive experimental verification further ensures the reliability and practicability of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0041] Figure 1This is a framework diagram of an optimal path planning algorithm for an electric robot according to the present invention. DETAILED DESCRIPTION

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0045] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0046] Example 1

[0047] An optimal path planning algorithm for an electric robot comprises the following steps:

[0048] S1, Environmental Modeling: Collect data through sensors, process point clouds and images, and generate three-dimensional maps that identify transmission lines and obstacles.

[0049] Specifically, S1 includes:

[0050] S11, uses sensors to collect point cloud data in a temporal environment, the data includes the robot's position, obstacle positions, and other environmental features;

[0051] S12, acquiring point cloud data from the sensor, and converting the point cloud data into a structured three-dimensional map using point cloud processing technology;

[0052] S13, determining the three-dimensional coordinates and geometric shape of the transmission line through measurement and data analysis, and marking it as a specific type of object;

[0053] S14, determining the location and size of the obstacle based on the sensor data and image analysis, and marking it as a three-dimensional shape on the three-dimensional map;

[0054] S15, dividing the three-dimensional map into grid units, each grid unit contains corresponding environmental information, setting a cost value for each grid unit according to the environmental information of each grid unit, and saving the processed three-dimensional map as an STL file, wherein the environmental information includes the spatial coordinates of the grid unit, whether there are obstacles in the grid unit, and the terrain in the grid unit.

[0055] S2, path planning algorithm: Combining the A-star algorithm and reinforcement learning method, the initial path planning is first performed through the A-star algorithm, and then the path is dynamically optimized and adjusted through the reinforcement learning method.

[0056] S3, Experiment and Verification: A large number of experiments are carried out in virtual environments and actual power inspection environments to verify the performance of the algorithm.

[0057] Example 2

[0058] An optimal path planning algorithm for an electric robot, based on Example 1, wherein initial path planning by using the A-star algorithm in S2 specifically comprises the following steps:

[0059] S21-1, define the starting grid unit P s and the end grid cell P g , add the starting point to the open list;

[0060] S21-2, take the grid unit with the smallest comprehensive cost from the open list as the current node, record the adjacent grid units of the current node as adjacent nodes, add the current node to the closed list, and for each adjacent node of the current node: if the adjacent node is impassable or is already in the closed list, skip it; if the adjacent node is not in the open list, add it to the open list, calculate its total cost value, and set the current node as its parent node; if the adjacent node is already in the open list, but the new total cost value is smaller, update its total cost value, and set the current node as its parent node, and repeat the above process until the end is found or the open list is empty;

[0061] S21-3, from the end point P g Backtrack to its parent node until the starting point P s , and obtain the path with the minimum cost as the initial path.

[0062] Specifically, the total cost calculation formula is:

[0063] f(n)=g(n)+h(n)

[0064] Among them, g(n) is the actual cost from the starting point to the current node, which is calculated by accumulating the cost value of each grid cell on the path, and h(n) is the estimated cost from the current node to the end point.

[0065] More specifically, the specific calculation formula for h(n) is:

[0066]

[0067] Among them, (x n ,y n ,z n ) is the current node coordinate, (x g ,y g ,z g ) are the end point coordinates.

[0068] In S2, the dynamic optimization and adjustment of the path by reinforcement learning method specifically includes the following steps:

[0069] S22-1, define an information vector containing information such as the current position of the robot, the position of obstacles, and the target position, define the operations that the robot can perform, including forward, backward, and turning actions, and define a reward function to give corresponding rewards based on whether the robot reaches the final target point, whether it successfully avoids new obstacles and returns to the initial path, whether it collides with obstacles, and the length of the path to avoid obstacles;

[0070] S22-2, establish a deep Q network, including a main network and a target network, the main network is used to select actions, and the target network is used to calculate the target Q value;

[0071] S22-3, training and adjusting main network parameters;

[0072] S22-4, validation in virtual and real environments.

[0073] Specifically, the reward function is:

[0074] R(s,a)=R g +R r +R o +R l

[0075] Among them, R g is the reward for the robot to reach the final target point. g =100, otherwise R g =0, R r The reward for the robot to avoid new obstacles and return to the initial path. When the robot starts to avoid new obstacles and its position coordinates coincide with the initial path at a certain moment, R r =50, otherwise R r =0, R o is the penalty for the robot colliding with an obstacle. o = -100, otherwise R o =0, Rl The robot is penalized for the length of the path it takes to avoid new obstacles. The longer the path, the greater the penalty value.

[0076] Specifically, R l The calculation formula is:

[0077] R l = -α·d(P t1 ,P t2 )

[0078] Among them, α is the penalty coefficient of path length, d(P t1 ,P t2 ) is the distance the robot moves from the beginning of avoiding the new obstacle to returning to the initial path.

[0079] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An optimal path planning algorithm for an electric robot, characterized in that: The following steps are involved: S1, Environmental Modeling: Collect data through sensors, process point clouds and images, and generate three-dimensional maps that identify transmission lines and obstacles; S2, path planning algorithm: Combining the A-star algorithm and reinforcement learning method, the initial path planning is first performed through the A-star algorithm, and then the path is dynamically optimized and adjusted through the reinforcement learning method; S3, Experiment and Verification: A large number of experiments are carried out in virtual environments and actual power inspection environments to verify the performance of the algorithm.

2. The optimal path planning algorithm for an electric robot according to claim 1, characterized in that: S1 specifically includes: S11, uses sensors to collect point cloud data in a temporal environment, the data includes the robot's position, obstacle positions, and other environmental features; S12, acquiring point cloud data from the sensor, and converting the point cloud data into a structured three-dimensional map using point cloud processing technology; S13, determining the three-dimensional coordinates and geometric shape of the transmission line through measurement and data analysis, and marking it as a specific type of object; S14, determining the location and size of the obstacle based on the sensor data and image analysis, and marking it as a three-dimensional shape on the three-dimensional map; S15, dividing the three-dimensional map into grid units, each grid unit contains corresponding environmental information, setting a cost value for each grid unit according to the environmental information of each grid unit, and saving the processed three-dimensional map as an STL file.

3. The optimal path planning algorithm for an electric robot according to claim 1, characterized in that: The environmental information in S15 includes the spatial coordinates of the grid unit, whether there are obstacles in the grid unit, and the terrain in the grid unit.

4. The optimal path planning algorithm for an electric robot according to claim 1, characterized in that: The initial path planning by the A-star algorithm in S2 specifically includes the following steps: S21-1, define the starting grid unit P s and the end grid cell P g , add the starting point to the open list; S21-2, take the grid unit with the smallest comprehensive cost from the open list as the current node, record the adjacent grid units of the current node as adjacent nodes, add the current node to the closed list, and for each adjacent node of the current node: if the adjacent node is impassable or is already in the closed list, skip it; if the adjacent node is not in the open list, add it to the open list, calculate its total cost value, and set the current node as its parent node; if the adjacent node is already in the open list, but the new total cost value is smaller, update its total cost value, and set the current node as its parent node, and repeat the above process until the end is found or the open list is empty; S21-3, from the end point P g Backtrack to its parent node until the starting point P s , and obtain the path with the minimum cost as the initial path.

5. The optimal path planning algorithm for an electric robot according to claim 4 is characterized in that , the total cost calculation formula is: f(n)=g(n)+h(n) Among them, g(n) is the actual cost from the starting point to the current node, which is calculated by accumulating the cost value of each grid cell on the path, and h(n) is the estimated cost from the current node to the end point.

6. The optimal path planning algorithm for an electric robot according to claim 5, characterized in that ,h(n) specific calculation formula is: Among them, (x n ,y n ,z n ) is the current node coordinate, (x g ,y g ,z g ) are the end point coordinates.

7. The optimal path planning algorithm for an electric robot according to claim 1, characterized in that: In S2, the dynamic optimization and adjustment of the path by reinforcement learning method specifically includes the following steps: S22-1, define an information vector containing information such as the current position of the robot, the position of obstacles, and the target position, define the operations that the robot can perform, including forward, backward, and turning actions, and define a reward function to give corresponding rewards based on whether the robot reaches the final target point, whether it successfully avoids new obstacles and returns to the initial path, whether it collides with obstacles, and the length of the path to avoid obstacles; S22-2, establish a deep Q network, including a main network and a target network, the main network is used to select actions, and the target network is used to calculate the target Q value; S22-3, training and adjusting main network parameters; S22-4, validation in virtual and real environments.

8. The optimal path planning algorithm for an electric robot according to claim 6, characterized in that: The reward function is specifically: R(s,a)=R g +R r +R o +R l Among them, R g is the reward for the robot to reach the final target point. g =100, otherwise R g =0, R r The reward for the robot to avoid the new obstacle and return to the initial path, R o is the penalty for the robot colliding with an obstacle. o = -100, otherwise R o =0, R l The robot is penalized for the length of the path it takes to avoid new obstacles. The longer the path, the greater the penalty value.

9. The optimal path planning algorithm for an electric robot according to claim 7, characterized in that: When the robot starts to avoid new obstacles and its position coordinates coincide with the initial path at a certain moment, R r =50, otherwise R r =0.

10. The optimal path planning algorithm for an electric robot according to claim 8, characterized in that: R l The calculation formula is: R l =-α·d(P t1 ,P t2 ) Among them, α is the penalty coefficient of path length, d(P t1 ,P t2 ) is the distance the robot moves from the beginning of avoiding the new obstacle to returning to the initial path.

Images