Mechanical arm obstacle avoidance planning method based on extended random tree connection algorithm

Abstract

The invention discloses a mechanical arm obstacle avoidance planning method based on an extended random tree connection algorithm. The mechanical arm obstacle avoidance planning method comprises the steps that S1, a starting point and a target point are initialized into root nodes of two random trees Tstart and Tgoal respectively; s2, in each iteration, generating a sampling point I in the working space; s3, a first node closest to the sampling point is searched in the random tree Tstart, a second new node is generated through extension according to the step length epsilon, and when a connecting line between the second node and a father node of the second node does not collide, the second node is added into the random tree Tstart; s4, searching a node 3 closest to the node 2 in the random tree Tgoal, generating a new node 4 through extension according to a step length epsilon, and adding the node 4 into the random tree Tgoal when a connecting line between the node 4 and a father node 3 of the node 4 is not collided; s5, after each round of iteration is finished, the roles of the two random trees are exchanged through Swap operation, and iteration is repeated until the distance between the second node and the fourth node is smaller than the extension step length epsilon or the maximum number of iterations is reached.

Description

Technical Field

[0001] This application relates to the field of robotic arm path planning technology, specifically to a robotic arm obstacle avoidance planning method based on an extended random tree connection algorithm. Background Technology

[0002] With industrial development, the trend of robots replacing manual labor has gradually spread to various industries. The widespread application of robotic arms has also led to in-depth research on robotic arms, moving towards automation, high speed, intelligence, and precision. While the structural design and posture control technologies of traditional rigid robotic arms are relatively mature, their inherent rigidity limits their ability to operate in unstructured, complex, and confined environments. Continuous robotic arms, with their high flexibility and multiple degrees of freedom, offer unparalleled advantages over traditional rigid robotic arms in narrow and complex working spaces, becoming a new avenue for robotic arm research. Furthermore, the research and development of these continuous robotic arms provides safer, more reliable, and intelligent options for industries such as rubble rescue, industrial manufacturing, medical testing, and aerospace. However, research on continuous robotic arms is still in its early stages, with very little research focusing on achieving autonomous obstacle avoidance. Therefore, it is necessary to explore and research autonomous obstacle avoidance in continuous robotic arms.

[0003] Application content

[0004] The purpose of this application is to provide a robotic arm obstacle avoidance planning method based on the extended random tree connection algorithm, and the specific technical solution is as follows:

[0005] A robotic arm obstacle avoidance planning method based on extended random tree connection algorithm, including: S1, setting the starting point P... start and target point P goal Initialize them as two random trees T respectively start and T goal The root node; S2, in each iteration, generate sampling points P in the workspace. rand S3, in the random tree T start Find the distance to sampling point P rand The nearest node And generate new nodes by expanding with a step size ε. When node with its parent node When there is no collision between the connections, the nodes will be... Add to random tree T start S4, in the random tree T goal Find distance nodes The nearest node And generate new nodes by expanding with a step size ε. When node with its parent node When there is no collision between the connections, the nodes will be... Add to random tree T goal S5. After each iteration, the roles of the two random trees are swapped using a Swap operation to allow bidirectional expansion to alternate, repeating the iteration until the node is reached. and nodes The distance between them is less than the expansion step size ε or the maximum number of iterations is reached.

[0006] In S2, the bias of sampling is controlled by bias_factor during sampling, which is used to guide sampling with a certain probability and converge more quickly when approaching the target.

[0007] In S3 and S4, a new node P is added. new Then, at node P new Search for an existing set of nodes within the neighborhood radius and compute the value of node P within that set. j Starting point P start and new node P new The path cost between the starting point P and P is... start To node P j To the new node P new The path cost is less than the starting point P. start To the new node P new When calculating the path cost, node P j Reset to new node P new The parent node is used to achieve local path optimization.

[0008] In S3 and S4, during the expansion process, the expansion step size ε gradually decreases as the number of iterations increases, in order to improve path accuracy.

[0009] S5 and nodes When the distance between them is less than the expansion step size ε, the starting point P start and target point P goal The connection was successful. The two path segments were extracted and smoothed to obtain the complete path.

[0010] In S2, when the bias of sampling is controlled by bias_factor, the expression is:

[0011] ,

[0012] in, Generate random sampling points P rand The bias_factor controls the sampling bias towards the target.

[0013] The advantages of this application lie in its improved search efficiency and reduced path length through a target-biased sampling strategy and dynamic step size adjustment. Furthermore, as the target point approaches, the sampling points gradually converge towards it, and the step size gradually decreases, thus improving search accuracy. Simultaneously, local reconnection expands and refines the branches of adjacent nodes, reducing path cost and achieving local optima, thereby enhancing global search capabilities. Finally, path smoothing techniques are employed to improve path smoothness and stability. In summary, the proposed algorithm significantly improves search speed, search time, and path quality, making it suitable for path planning in complex environments.

[0014] Instruction manual illustrations

[0015] Figure 1 This is a flowchart illustrating the application process.

[0016] Figure 2 This is a schematic diagram of path planning under general obstacle conditions in this application;

[0017] Figure 3 This is a schematic diagram of path planning in a complex obstacle environment in this application. Specific Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to specific embodiments and accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of this application. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0019] like Figure 1 As shown:

[0020] A robotic arm obstacle avoidance planning method based on an extended random tree connection algorithm includes:

[0021] S1, Set the starting point P start and target point P goal Initialize them as two random trees T respectively start and T goal The root node.

[0022] S2. In each iteration, generate sampling points P in the workspace. rand During sampling, the bias is controlled by `bias_factor` to guide the sampling with a certain probability and to converge more quickly when approaching the target. The expression for controlling the sampling bias using `bias_factor` is:

[0023] ,

[0024] in, Generate random sampling points P rand The bias_factor controls the sampling bias towards the target.

[0025] S3, in the random tree T start Find the distance to sampling point P rand The nearest node And generate new nodes by expanding with a step size ε. When node with its parent node When there is no collision between the connections, the nodes will be... Add to random tree T start .

[0026] S4, in the random tree T goal Find distance nodes The nearest node And generate new nodes by expanding with a step size ε. When node with its parent node When there is no collision between the connections, the nodes will be... Add to random tree T goal .

[0027] In S3 and S4, a new node P is added. new Then, at node P new Search for an existing set of nodes within the neighborhood radius and compute the value of node P within that set. j Starting point P start and new node P new The path cost between the starting point P and P is... start To node P j To the new node P new The path cost is less than the starting point P. start To the new node P new When calculating the path cost, node P j Reset to new node P new The parent node is used to achieve local path optimization. During expansion, the expansion step size ε gradually decreases with the increase of the number of iterations to improve path accuracy.

[0028] S5. After each iteration, the roles of the two random trees are swapped using a Swap operation to allow bidirectional expansion to proceed alternately. This process is repeated (steps S2-S4 are repeated) until the node is reached. and nodes If the distance between them is less than the expansion step size ε or the maximum number of iterations is reached, the roles of the two random trees are swapped through a Swap operation after each round of iterations, so that bidirectional expansion can be carried out alternately. and nodes When the distance between them is less than the expansion step size ε, the starting point P startand target point P goal The connection was successful. The two path segments were extracted and smoothed to obtain the complete path.

[0029] To make this application easier to understand, the following analysis will focus on its specific implementation.

[0030] To more intuitively analyze the performance of path planning algorithms, we simulated and analyzed the performance of four path planning algorithms. To ensure the representativeness of the simulation results, two typical environments were constructed: a general obstacle environment and a complex obstacle environment. Each algorithm underwent 30 independent simulations, and the average of these 30 simulations was taken as the performance index, including the number of iterations, path search time, and path length. The simulations were performed using MATLAB, with a 120mm × 120mm two-dimensional plane, a starting point of (0,0), an ending point of (100,100), a maximum number of iterations of 1000, a step size of 5, and a distance threshold of 5.

[0031] algorithm Average number of iterations Average path length (mm) Average planning time (s) RRT 105 158.57 1.0723 RRT-Connect 172 149.63 1.1750 RRT* 372 155.52 3.5944 RRT-Connect-Improved 64 140.01 0.8768

[0032] Table 1

[0033] from and Figure 2 As can be seen from the results, after 30 simulations under general obstacle conditions, the path planning algorithm (RRT-Connect-Improved) provided in this application has fewer average iterations, path length, and planning time than the other three algorithms. The improved algorithm reduces the average number of iterations by 62.8%, the average path length by 6.4%, and the average planning time by 25.4% compared to the traditional RRT-Connect algorithm. The path length of the path planning algorithm (RRT-Connect-Improved) provided in this application is slightly shorter than that of RRT-Connect, but shorter than the other two algorithms, indicating a significant improvement in path optimization.

[0034] like Figure 3 As shown, more obstacles were added to further verify the effectiveness of the path planning algorithm (RRT-Connect-Improved) provided in this application.

[0035] algorithm Average number of iterations Average path length (mm) Average planning time (s) RRT 227 166.60 1.4003 RRT-Connect 228 155.44 0.9028 RRT* 236 160.20 1.6584 RRT-Connect-Imp 119 150.39 0.7524

[0036] Table 2

[0037] according to Figure 3As shown in Table 2, the path planning algorithm (RRT-Connect-Improved) provided in this application exhibits good stability in complex obstacle environments. It has the fewest average number of iterations, path length, and planning time. The average number of iterations of the improved algorithm is reduced by 47.8%, the average path length is reduced by 3.2%, and the average planning time is reduced by 16.7% compared with the RRT-Connect algorithm. This once again verifies the performance of the improved algorithm in complex obstacle environments of different degrees, demonstrating its superiority and efficiency in complex environments.

Claims

1. A robotic arm obstacle avoidance planning method based on an extended random tree connection algorithm, characterized in that, include: S1, Set the starting point P start and target point P goal Initialize them as two random trees T respectively start and T goal The root node; S2. In each iteration, generate sampling points P in the workspace. rand ; S3, in the random tree T start Find the distance to sampling point P in the middle rand The nearest node And generate new nodes by expanding with a step size ε. When node with its parent node When there is no collision between the connections, the nodes will be... Add to random tree T start ; S4, in the random tree T goal Find distance nodes The nearest node And generate new nodes by expanding with a step size ε. When node with its parent node When there is no collision between the connections, the nodes will be... Add to random tree T goal ; S5. After each iteration, the roles of the two random trees are swapped using a Swap operation to allow bidirectional expansion to alternate, repeating the iteration until the node is reached. and nodes The distance between them is less than the expansion step size ε or the maximum number of iterations is reached.

2. The obstacle avoidance planning method for robotic arms based on the extended random tree connection algorithm as described in claim 1, characterized in that, In S2, the bias of sampling is controlled by bias_factor during sampling, which is used to guide sampling with a certain probability and converge more quickly when approaching the target.

3. The obstacle avoidance planning method for robotic arms based on the extended random tree connection algorithm as described in claim 1, characterized in that, In S3 and S4, a new node P is expanded. new Then, at node P new Search for an existing set of nodes within the neighborhood radius and compute the value of node P within that set. j Starting point P start and new node P new The path cost between the starting point P and P is... start To node P j To the new node P new The path cost is less than the starting point P. start To the new node P new When calculating the path cost, node P j Reset to new node P new The parent node is used to achieve local path optimization.

4. The obstacle avoidance planning method for robotic arms based on the extended random tree connection algorithm as described in claim 1, characterized in that, In S3 and S4, during the expansion process, the expansion step size ε gradually decreases as the number of iterations increases, in order to improve path accuracy.

5. The obstacle avoidance planning method for robotic arms based on the extended random tree connection algorithm as described in claim 1, characterized in that, In S5 and nodes When the distance between them is less than the expansion step size ε, the starting point P start and target point P goal The connection was successful. The two path segments were extracted and smoothed to obtain the complete path.

6. The obstacle avoidance planning method for a robotic arm based on the extended random tree connection algorithm as described in claim 2, characterized in that, In S2, when the bias of sampling is controlled by bias_factor, the expression is: ， in, Generate random sampling points P rand The bias_factor controls the sampling bias towards the target.

Images