A Robotic Arm Path Planning Method Based on an Improved RRT* Algorithm
By improving the greedy strategy and target bias strategy of the RRT* algorithm, optimizing the path search termination condition and removing redundant points, the speed and redundancy problems of the RRT* algorithm in path planning are solved, achieving faster search and shorter paths.
Patent Information
- Application Number
- CN202410186778.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-20
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2044-02-20
AI Technical Summary
Existing RRT* algorithms suffer from slow search speed, long convergence time, and numerous redundant points in path planning, especially in complex environments.
By introducing a greedy strategy to determine whether the starting point and the target point can be reached directly, the termination condition of the path search is optimized. A probability-based target bias strategy is introduced to guide the direction of node expansion, and the path is optimized to remove redundant points.
It improves the search speed of path planning, shortens the convergence time, reduces path length and redundant points, and improves the efficiency of path planning.
Smart Images

Figure CN117841001B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of robotic arm path planning technology, and particularly relates to a robotic arm path planning method based on an improved RRT* algorithm. Background Technology
[0002] Path planning is a critical problem in automated systems, involving finding the optimal path or a path that satisfies specific constraints within a given environment. In the aerospace field, manufactured blades have complex structures and require surface machining. Currently, these complex curved surfaces are primarily machined manually by skilled and experienced workers. However, manual machining suffers from low production efficiency, difficulty in guaranteeing the surface quality of complex free-form surfaces, and the dust generated during processing can impact worker health. Furthermore, due to the complex structures of blades, blade roots, and blade crowns, path planning is necessary when machining at different locations. Without precise surface path planning, machining efficiency and quality will be reduced. Manual machining is not practical for various machining situations, cannot be tailored to specific engineering needs, and the planning methods are relatively simplistic and inflexible. Therefore, a more precise path planning method is needed to adapt to the needs of a wider range of engineering projects.
[0003] The RRT* (Rapidly-exploring Random Tree Star) algorithm is a stochastic exploration algorithm for path planning (e.g., ...). Figure 1 (As shown). It is an improvement on the RRT algorithm, introducing additional optimization steps to reduce the cost of generating paths. It is a significant improvement over the RRT algorithm. However, since the RRT* algorithm still randomly selects points in the environment, it still has problems such as slow search speed, long convergence time, and many redundant points in the path.
[0004] To address the aforementioned issues, the paper "A continuous RRT*-based path planning method for non-holonomic mobile robots using B-spline curves[J]. Journal of Ambient Intelligence and Humanized Computing, 2023, 14(7): 8693-8702" proposes a combined method based on RRT* and B-spline curves to smooth paths generated by the RRT* algorithm. However, this paper only smooths the paths obtained after the search is completed and does not solve the problems of slow search speed, long convergence time, and many redundant points in the path of the RRT* algorithm itself. Summary of the Invention
[0005] The purpose of this invention is to solve the problems existing in the prior art and provide a robotic arm path planning method based on the improved RRT* algorithm.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0007] A robotic arm path planning method based on an improved RRT* algorithm includes the following steps:
[0008] Step 1: Construct the workspace of the robotic arm, initialize the random tree, and determine the starting and target points of the motion;
[0009] Step 2: Determine if the starting point and the target point are directly reachable. If they are, connect the starting point and the target point to obtain a collision-free path from the starting point to the target point. Otherwise, set a step size and expand the nodes from the starting point towards the target point until the last collision-free node q is reached. new1 The last node q that does not collide new1 As the parent node;
[0010] Step 3: Expand a new node from its parent node;
[0011] Step 4: Determine if the path search termination condition is met. If not, set the new node as the parent node and return to Step 3; otherwise, proceed to the next step.
[0012] The termination condition for path search is L≤R, where L is the distance between the last new node and the target point, and R is the radius of the largest circular collision-free region centered on the target point (or the largest fan-shaped collision-free region if it is at the environment boundary).
[0013] Step 5: Process all points to obtain a collision-free path from the starting point to the target point.
[0014] The improvements of this invention to the RRT* algorithm include: ① using a greedy strategy to determine whether the starting point and the target point are directly reachable; if they are, the search is completed; otherwise, nodes are expanded from the starting point towards the target point; ② optimizing the termination condition for path search. These two improvements enable this invention to solve the problems of slow search speed and long convergence time in the RRT* algorithm, because this invention reduces some node expansion and guides the direction of node expansion.
[0015] As a preferred technical solution:
[0016] In the above-mentioned path planning method for a manipulator based on the improved RRT* algorithm, in step 2, the process of judging whether the start point and the target point can directly reach each other is: directly connect the start point to the target point, and perform obstacle collision detection. If no collision occurs, the direct reach is achievable; otherwise, the direct reach is not achievable.
[0017] In the above-mentioned path planning method for a manipulator based on the improved RRT* algorithm, in step 3, the process of expanding a new node from the parent node is: set a target deviation probability threshold Ptarget, randomly generate a probability value P where P∈[0,1], judge whether P<Ptarget is true. If yes, randomly generate a sampling point, and expand a new node from the parent node in the direction of the sampling point; otherwise, expand a new node from the parent node in the direction of the target point.
[0018] The improvements of the present invention to the RRT* algorithm further include: ③ introducing a probability-based target bias strategy during the node expansion process to guide the growth direction of the random tree nodes; the presence of improvement ③ can reduce the blindness of search and accelerate the search speed.
[0019] In the above-mentioned path planning method for a manipulator based on the improved RRT* algorithm, Ptarget=0.6.
[0020] In the above-mentioned path planning method for a manipulator based on the improved RRT* algorithm, in step 4, the process of determining the value of R is as follows:
[0021] (i) Let R be equal to the step size;
[0022] (ii) Taking the target point as the center, expand outward with R as the radius;
[0023] (iii) Judge whether a collision occurs. If yes, output R; otherwise, let R=R+ΔR where ΔR∈[0,R], and return to step (ii).
[0024] In the above-mentioned path planning method for a manipulator based on the improved RRT* algorithm, in step 5, the specific process of processing all points to obtain a collision-free path from the start point to the target point is: optimize the path, remove redundant points in the path, and obtain a collision-free path from the start point to the target point.
[0025] The improvements of the present invention to the RRT* algorithm further include: ④ optimizing the path and removing redundant points in the path; the presence of improvement ④ can solve the problem of excessive redundant path points existing in the RRT* algorithm.
[0026] In the above-mentioned path planning method for a manipulator based on the improved RRT* algorithm, step 5 is specifically as follows:
[0027] (a) Number each node in ascending order of appearance;
[0028] (b) Use the target point as the starting point;
[0029] (c) Connect the starting point to each node, perform obstacle collision detection on each line, record the nodes that do not collide with obstacles, and find the node with the largest number that does not collide with obstacles.
[0030] (d) Connect the target point, the node with the largest number that did not collide with the obstacle obtained in step (c) in each loop, and the starting point in sequence. Determine whether a collision-free path from the starting point to the target point can be obtained. If so, end the loop; otherwise, take the node with the largest number that did not collide with the obstacle as the starting point and return to step (c).
[0031] Beneficial effects:
[0032] (1) This invention uses a greedy strategy to determine whether the starting point and the target point can be reached directly, optimizes the termination condition of the path search, reduces the expansion of some nodes, and guides the direction of node expansion, thus solving the problems of slow search speed and long convergence time of the RRT* algorithm.
[0033] (2) This invention introduces a probability-based target bias strategy to guide the growth direction of random tree nodes during the node expansion process, which can reduce the blindness of the search, speed up the search, optimize the path, remove redundant points in the path, shorten the path length, and improve the efficiency of path planning. Attached Figure Description
[0034] Figure 1 This is a schematic diagram of the RRT* algorithm in the prior art;
[0035] Figure 2 This is a schematic diagram illustrating the use of a greedy strategy to connect the starting point and the target point in this invention;
[0036] Figure 3 This is a schematic diagram illustrating the optimization of the path termination conditions in this invention.
[0037] Figure 4 This is a schematic diagram illustrating the path optimization and redundant point removal of the present invention.
[0038] Figure 5 The image shows the path planning experiment obtained using the RRT* algorithm in a simple two-dimensional environment.
[0039] Figure 6 The image shows the path planning experiment obtained using the improved RRT* algorithm in a simple two-dimensional environment.
[0040] Figure 7This is an experimental diagram of path planning obtained using the RRT* algorithm in a two-dimensional complex environment;
[0041] Figure 8 The image shows the path planning experiment obtained using the improved RRT* algorithm in a two-dimensional complex environment.
[0042] Figure 9 The image shows an experimental diagram of the robotic arm path planning obtained in a 3D environment using the RRT* algorithm.
[0043] Figure 10 The experimental diagram of the robotic arm path planning obtained in a three-dimensional environment using the improved RRT* algorithm;
[0044] Figure 11 The flowchart shows the robotic arm path planning method based on the improved RRT* algorithm of the present invention.
[0045] Figure 12 The image shows the path planning experiment obtained using the PRRT* algorithm in a simple two-dimensional environment.
[0046] Figure 13 This is an experimental diagram of path planning obtained using the PRRT* algorithm in a complex two-dimensional environment;
[0047] Figure 14 This is an experimental diagram of the robotic arm path planning obtained in a 3D environment using the PRRT* algorithm;
[0048] Figure 15 The figure shows a comparison of the path planning results obtained by the three algorithms in a three-dimensional environment. Detailed Implementation
[0049] The present invention will be further described below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.
[0050] A robotic arm path planning method based on an improved RRT* algorithm, such as... Figure 11 As shown, it includes the following steps:
[0051] Step 1: Construct the workspace of the robotic arm, initialize the random tree, and determine the starting and target points of the motion;
[0052] Step 2: As Figure 2As shown, directly connect the starting point to the target point, and perform obstacle collision detection. If no collision occurs, connect the starting point and the target point to obtain a collision-free path from the starting point to the target point; otherwise, after setting the step size, expand nodes from the starting point towards the target point until the last node q that does not collide is obtained new1 , the last non-colliding node q new1 is taken as the parent node;
[0053] Step 3: Set a target deviation probability threshold Ptarget, randomly generate a probability value P where P∈[0,1], and determine whether P<Ptarget (Ptarget=0.6) is true. If yes, randomly generate a sampling point, and expand a new node from the parent node towards the sampling point; otherwise, expand a new node from the parent node towards the target point;
[0054] Step 4: As shown Figure 3 , determine whether the termination condition for path search is satisfied. If not, take the new node as the parent node and return to Step 3; otherwise, proceed to the next step;
[0055] The termination condition for path search is L≤R, where L is the distance between the last new node that appears and the target point, and R is the radius of the maximum circular collision-free area centered on the target point;
[0056] The process of determining the value of R is as follows:
[0057] (i) Let R be equal to the step size;
[0058] (ii) Centered on the target point, expand outward with R as the radius;
[0059] (iii) Determine whether a collision occurs. If yes, output R; otherwise, let R=R+ΔR where ΔR∈[0,R], and return to step (ii);
[0060] Step 5: Process all points to obtain a collision-free path from the starting point to the target point, as shown Figure 4 , the details are as follows:
[0061] (a) Number each node in ascending order according to the reverse order of their appearance, that is, from the latest appeared node to the earliest appeared node;
[0062] (b) Take the target point as the starting point;
[0063] (c) Connect the starting point to each node respectively, perform obstacle collision detection on each connecting line, record the nodes that do not collide with obstacles, and find the node with the largest number that does not collide with obstacles;
[0064] (d) Connect the target point, the node with the largest number that did not collide with the obstacle obtained in step (c) in each loop, and the starting point in sequence. Determine whether a collision-free path from the starting point to the target point can be obtained. If so, end the loop; otherwise, take the node with the largest number that did not collide with the obstacle as the starting point and return to step (c).
[0065] This invention has the advantages of fast search speed, short average path length, and low number of redundant points. To support this point, the robotic arm path planning method based on the improved RRT* algorithm described above was used to perform path planning from the starting point to the target point in both two-dimensional and three-dimensional environments. The results are as follows:
[0066] A robotic arm path planning method based on an improved RRT* algorithm is used in simple two-dimensional environments (such as...). Figure 6 The average search time for path planning (as shown) is 2.152861E-02s, the average path length is 6.222540E+02mm, the average number of sampling points is 0, and the average number of path points is 2.
[0067] A robotic arm path planning method based on an improved RRT* algorithm is used in complex two-dimensional environments (such as...). Figure 8 The average search time for path planning (as shown) is 1.339917E+01s, the average path length is 6.372254E+02mm, the average number of sampling points is 332.7, and the average number of path points is 3.82.
[0068] A robotic arm path planning method based on an improved RRT* algorithm is adopted in a three-dimensional environment (such as...). Figure 10 and Figure 15 The average search time for path planning (as shown) is 7.388160E-01s, the average path length is 3.868503E+02mm, the average number of sampling points is 102.64, and the average number of path points is 5.96.
[0069] To demonstrate that this invention is superior to the prior art, the improved RRT* algorithm of this invention was replaced with the RRT* algorithm or the PRRT* algorithm (the target bias strategy of this invention was added to the RRT* algorithm, which is the PRRT* algorithm), and path planning from the starting point to the target point was performed in two-dimensional and three-dimensional environments, respectively. The results are as follows:
[0070] Robotic arm path planning using the RRT* algorithm in simple two-dimensional environments (such as...) Figure 5 The average search time for path planning (as shown) is 1.155844E+02s, the average path length is 6.403177E+02mm, the average number of sampling points is 1907.8, and the average number of path points is 10.6.
[0071] The robotic arm path planning method using the RRT* algorithm is applicable to complex two-dimensional environments (such as...). Figure 7 The average search time for path planning (as shown) is 6.227293E+01s, the average path length is 6.507086E+02mm, the average number of sampling points is 1462.68, and the average number of path points is 10.94.
[0072] The robotic arm path planning method using the RRT* algorithm in a three-dimensional environment (such as...) Figure 9 and Figure 15 The average search time for path planning (as shown) is 1.001153E+01s, the average path length is 4.085839E+02mm, the average number of sampling points is 2252.78, and the average number of path points is 10.48.
[0073] The robotic arm path planning method using the PRRT* algorithm is applied in simple two-dimensional environments (such as...). Figure 12 The average search time for path planning (as shown) is 1.926396E+02s, the average path length is 6.438476E+02mm, the average number of sampling points is 2427.64, and the average number of path points is 11.44.
[0074] The robotic arm path planning method using the PRRT* algorithm is applicable to complex two-dimensional environments (such as...). Figure 13 The average search time for path planning (as shown) is 1.955160E+02s, the average path length is 6.593034E+02mm, the average number of sampling points is 2241.8, and the average number of path points is 11.64.
[0075] The robotic arm path planning method using the PRRT* algorithm in a three-dimensional environment (e.g., Figure 14 and Figure 15 The average search time for path planning (as shown) is 8.391033E-01s, the average path length is 3.880759E+02mm, the average number of sampling points is 120.64, and the average number of path points is 7.64.
[0076] The above results show that, in a two-dimensional environment, the improved RRT* algorithm of this invention reduces the average search time by 78.49% and 93.15% compared to the RRT* and PRRT* algorithms, and reduces the average path length by 2.1% and 3.35%, respectively. Figure 15 As shown, in a three-dimensional environment, the improved RRT* algorithm of this invention (i.e., ImRRT*) reduces the average search time by 92.63% and 11.96% compared to the RRT* and PRRT* algorithms, and reduces the average path length by 5.4% and 0.32% respectively. These data are sufficient to prove that the improved RRT* algorithm of this invention has the advantages of fast search speed, short average path length, and low number of redundant points.
[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A robotic arm path planning method based on an improved RRT* algorithm, characterized in that, comprising the following steps: Step 1: Construct the working space of the manipulator, initialize a random tree, and determine the starting point and the target point of the motion; Step 2: Determine whether the starting point and the target point can be directly reached; if direct reach is possible, connect the starting point and the target point to obtain a collision-free path from the starting point to the target point; Conversely, after setting the step size, the nodes are expanded from the starting point towards the target point until the last node q that does not collide. new1 The last node q that does not collide new1 As the parent node; Step 3: Set a target deviation probability threshold Ptarget, randomly generate a probability value P where P∈[0,1], determine whether P<Ptarget is satisfied; if yes, randomly generate a sampling point, and expand a new node from the parent node in the direction of the sampling point; otherwise, expand a new node from the parent node in the direction of the target point; Step 4: Determine whether the termination condition for path search is satisfied; if not, take the new node as the parent node and return to Step 3; otherwise, proceed to the next step; The termination condition for path search is L≤R, where L is the distance between the last occurring new node and the target point, and R is the radius of the largest circular collision-free area centered on the target point; Step 5: Optimize the path, remove redundant points in the path, and obtain a collision-free path from the starting point to the target point.
2. The robotic arm path planning method based on the improved RRT* algorithm according to claim 1, characterized in that, In Step 2, the process of determining whether the starting point and the target point can be directly reached is: directly connect the starting point to the target point, and perform obstacle collision detection; if no collision occurs, direct reach is possible; otherwise, direct reach is not possible.
3. The robotic arm path planning method based on the improved RRT* algorithm according to claim 1, characterized in that, Ptarget=0.
6.
4. The robotic arm path planning method based on the improved RRT* algorithm according to claim 1, characterized in that, In Step 4, the process of determining the value of R is as follows: (i) Set R equal to the step length; (ii) Expand outward with the target point as the center and R as the radius; (iii) Determine whether a collision occurs; if yes, output R; otherwise, set R=R+ΔR where ΔR∈[0,R], and return to step (ii).
5. A robotic arm path planning method based on an improved RRT* algorithm according to claim 1, characterized in that, Step 5 is specifically as follows: (a) Number each node in ascending order according to the sequence from the latest to the earliest occurrence; (b) Take the target point as the starting point; (c) Connect the starting point to each node respectively, perform obstacle collision detection on each connecting line, record the nodes that do not collide with obstacles, and find the node with the largest number that does not collide with any obstacle; (d) Connect the target point, the node with the largest number that does not collide with any obstacle obtained in step (c) in each cycle, and the starting point in sequence, determine whether a collision-free path from the starting point to the target point can be obtained; if yes, the process ends; otherwise, take the node with the largest number that does not collide with any obstacle as the new starting point, and return to step (c).
Citation Information
Patent Citations
Novel bidirectional RRT* method suitable for mechanical arm joint space path planning
CN114310904A
Motion path planning method and device based on RRT algorithm, terminal and medium
CN116518965A
