Ship cabin welding mechanical arm obstacle avoidance motion path planning method

Abstract

The invention discloses a ship cabin welding mechanical arm obstacle avoidance motion path planning method. The method comprises the steps that starting points and target points of two random trees are set; generating a random tree through bidirectional growth; a sampling point xrand is generated, and target bias is carried out; a nearest node xnear is searched for; calculating the resultant force Ft of the attractive force and the repulsive force borne by each nearest node xnear; expanding from the nearest node xnew to the direction at a fixed step length s to generate a new node xnew; collision detection; and iterating repeatedly to generate a preliminary path. A target bias strategy is added into a bidirectional RRT * algorithm, an artificial potential field is improved to plan the tail end movement track of the welding machine, the algorithm combines attraction force and repulsive force in the path planning process, obstacles can be effectively bypassed, the search process is more concentrated in a target area, the connectivity and planning quality of a path are enhanced, and the path planning efficiency is improved. The problems that an RRT * algorithm is uneven in sampling, too many in redundant points, too many in iteration times and too long in path are solved, the algorithm searching efficiency is improved, and the obstacle avoidance motion of the welding mechanical arm is more efficient and safer.

Description

Technical Field

[0001] The invention relates to an obstacle avoidance motion path planning method for a ship cabin welding mechanical arm, and belongs to the technical field of path planning. Background Art

[0002] In the shipbuilding industry, there are many narrow spaces where welding operations are required. Traditional manual welding operations have many disadvantages and cannot meet the process requirements. With the development of the robot industry, robotic arms are often used to assist production in today's shipbuilding industry. Robotic arms need to perform motion planning when performing tasks, and path planning is a key component of robotic arm motion planning. During ship welding operations, welding robotic arms must have flexible task execution capabilities and autonomous and efficient path planning functions to avoid obstacles and accurately reach the designated location. This allows a collision-free path from the starting position to the target position to be quickly selected in a complex obstacle environment.

[0003] The Rapidly-exploring Random Tree (RRT) algorithm is a widely used method in robot path planning. Based on a random sampling strategy, the algorithm searches in a wide space to find feasible solutions, thereby increasing the probability of finding the global optimal solution. However, due to its reliance on randomness, the RRT algorithm is prone to uneven distribution of sampling points in the search space. This unevenness can easily cause discontinuity or unevenness in the planned path, which in turn affects the quality of the path. A low-quality path can easily lead to unnecessary vibration problems when the robot moves, which not only affects the accuracy of the operation, but also poses potential risks to the robot and its environment.

[0004] The patent with publication number CN 116852367 A discloses a six-axis robot obstacle avoidance path planning method based on an improved RRT* algorithm. This method introduces a target bias and dual-tree expansion strategy to address the problems of the original algorithm's large search space, long search time, complex calculations, and uneven paths. It simultaneously expands from the starting point and the target point, combines the strategy of shortening the path and removing redundant nodes, and optimizes the path by B-spline fitting. This method effectively reduces the search time and complexity, achieves path shortening and smoothing, and ensures that the end of the robot arm can avoid obstacles efficiently and safely. However, its algorithm will cause problems such as many redundant points, many iterations, and long paths, and this method only uses B-splines to fit a smooth path, ignoring the pruning of the initially generated path to further shorten the path. Summary of the invention

[0005] Purpose of the invention: In view of the shortcomings of the prior art, the present invention provides a method for obstacle avoidance motion path planning of a ship cabin welding robot arm. The present invention adds a target bias strategy and an improved artificial potential field mechanism to improve path search efficiency and optimize path quality.

[0006] Technical solution: A method for obstacle avoidance motion path planning of a ship cabin welding robot arm, comprising the following steps:

[0007] Step 1: Set the starting point and target point of the two random trees: Determine the starting point x of the random tree T1 according to the starting point of the robot arm start1 and the target point x of random tree T2 goal2 , determine the starting point x of random tree T2 according to the welding end point of the robot arm start2 and the target point x of random tree T1 goal1 ;

[0008] Step 2: Generate random trees with bidirectional growth: two starting points x start1 and x start2 and their corresponding target points x goal1 and x goal2 Perform bidirectional growth and generate two random trees T respectively 1 and T 2 ;

[0009] Bidirectional growth means exploring and expanding outward from the starting point and the target point respectively until the two trees meet and the planning algorithm converges, which improves the efficiency of finding the path.

[0010] Step 3: Generate sampling point x rand And perform target bias on it: Generate random points in the workspace, and according to the preset probability P, move the random points to the target point x goal1 Or the target point x goal2 Bias, respectively obtain the sampling point x under the preset probability P rand ;

[0011] Step 4: Find the nearest node x near : For any random tree T 1 or T 2 , find the distance from each sampling point x in step three rand The nearest node x near , that is, with the sampling point x rand The node x with the smallest distance near .

[0012] Step 5: Use the improved artificial potential field function to calculate the nearest node x near Calculate the combined force F of its gravitational and repulsive forces t ;

[0013] Step 6: According to the calculated resultant force Ft Direction, from the nearest node x near Expand in this direction with a fixed step size s to generate a new node x new ;

[0014] Step 7: Generate new node x new Perform collision detection, that is, determine the new node x new Check whether the connection path with its parent node interferes with obstacles. If a collision occurs, abandon the current expansion and return to step 3 to resample and expand. If no collision occurs, optimize the new node x. new and the nearest node x near The path length between them goes to step eight;

[0015] Step 8. Set the maximum number of iterations n max , repeat steps 3 to 7, and check in each iteration whether the two random trees T1 and T2 are successfully grown and the path connection between them is completed. If the maximum number of iterations n max Inside, two random trees T 1 and T 2 The paths are successfully connected, that is, the two random trees start from their respective starting points x start1 and x start2 Towards their respective target points x goa l1 and x goal2 When they meet and connect during the growth process, a preliminary path is generated and sampling stops; if the maximum number of iterations n is not reached, max Inner join the two trees and return to step three.

[0016] Preferably, the step three is specifically as follows:

[0017] Since the path search of the RRT* algorithm in a three-dimensional complex environment has the problems of high randomness and long search time, a dual-target bias strategy is adopted as formula (1) for random sampling:

[0018]

[0019] Where rand is a random number between 0 and 1, p is the set target bias threshold, and size(map) is the size of the map;

[0020] When the value of rand is greater than the set target bias threshold, the random point is used as the sampling point x rand ; Otherwise, the target point x goal1 and x goal2 As the sampling points x of the respective random trees rand .

[0021] Preferably, the step five is specifically as follows:

[0022] Target point x goal For the nearest node x near Apply attraction F ai , while sampling point x rand Also for the nearest node x near Apply attraction F aj , at the nearest node x near The surrounding obstacles exert a repulsive force F on it ri , comprehensive attraction F ai and the repulsive force F aj , determine the final direction F of node expansion t :

[0023] The artificial potential field includes the gravitational field function U att As shown in formula (2), the repulsive field function U rep As shown in formula (3), the resultant potential field function is shown in formula (4):

[0024]

[0025] U=U att +U rep (4)

[0026] Among them, K a is the gravitational factor; K r is the repulsion factor; d is the distance between the robot arm and the obstacle; d 0 is the range of the repulsive potential field of the obstacle; P r is the position coordinate of the end of the robot arm; P g is the target point position coordinate; a is the improvement factor, d(P r -P g ) is the relative distance between the end of the robot arm and the target point;

[0027] Each nearest node x near The combined force of attraction and repulsion is F t For: F t =F aj +F ri .

[0028] Preferably, in step seven, the new node x is optimized new and the nearest node x near The path length between is:

[0029] The new node x new Take any nearest node x near Join as parent node with the nearest node x near The corresponding random tree T 1 or T 2 In the new node x newSearch for random tree T within the radius r of the center 1 or T 2 In the node, select the node that can make the start1 and x start2 Go to the new node x respectively new The node x with the shortest path min , and the new node x new The parent node is updated to x min , to optimize the path.

[0030] The preferred options also include optimization of the initial path length and the number of turns, specifically:

[0031] Traverse the initial path in step eight except the starting point x start1 、Target point x goal1 Or starting point x start2 、Target point x goal2 Remove some intermediate nodes, check whether the preliminary path after removing the intermediate nodes has no collision with obstacles, if there is no collision, update the preliminary path, if there is a collision, restore the intermediate nodes, select other intermediate nodes to remove, until the preliminary path has no collision with obstacles, and complete the optimization of the preliminary path length and the number of turns.

[0032] The preferred option also includes optimizing the linearity of the preliminary path, specifically:

[0033] Except for the starting point x start1 、Target point x goal1 Or starting point x start2 、Target point x goal2 The path connection is performed between two non-adjacent intermediate nodes other than the first node and the second node is checked to see if the path has no collision with obstacles. If there is no collision, the nodes between the two non-adjacent intermediate nodes are removed and the preliminary path is updated. If a collision occurs, the path connection between the two non-adjacent intermediate nodes is cancelled and non-adjacent intermediate nodes are reselected for path connection until the connection path has no collision with obstacles and the optimization of the linearity of the preliminary path is completed.

[0034] The preferred option also includes optimizing the smoothness of the preliminary path, specifically:

[0035] Preliminary path smoothness pre-optimization:

[0036] From the target point x of the preliminary path goal1 Start with x start1 The intermediate nodes other than the target point x are connected one by one. goal1 If the path connecting to the intermediate node does not collide with the obstacle, remove the target point x. goal1 and all nodes between the intermediate node, and update the path. If the target point x goal1If the target point x collides with the connection path of the intermediate node, goal1 Connect with the previous intermediate node that does not collide, and repeat the above connection steps with the intermediate node that collides as the new target point until the connection path does not collide with the obstacle, completing the optimization of the initial smoothness of the path. This optimization method is the same as starting from the target point x of the initial path. goal2 Start with x start2 The preliminary path optimization method for connecting the intermediate nodes other than the first node one by one is the same;

[0037] Then, the B-spline curve is used to smoothly fit the path after the preliminary path smoothness is pre-optimized, specifically:

[0038]

[0039] Where P i is the control point of the curve, F i,k (t) is the k-order B-spline basis function.

[0040] Beneficial effects: The present invention plans the end motion trajectory of the welding machine by adding a target bias strategy and improving the artificial potential field to the bidirectional RRT* algorithm. The algorithm combines attraction and repulsion in the path planning process, which can effectively bypass obstacles and make the search process more focused on the target area, thereby enhancing the connectivity and planning quality of the path. The improved path planning algorithm solves the problems of uneven sampling, too many redundant points, too many iterations, and too long paths in the RRT* algorithm, and improves the search efficiency of the algorithm, thereby making the obstacle avoidance movement of the welding robot arm more efficient and safe. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0042] Figure 1 is a flow chart of the method of the present invention;

[0043] Figure 2 This is a diagram of the generation process of two random trees for the bidirectional RRT* algorithm;

[0044] Figure 3 This is a diagram of the preliminary path optimization process of this embodiment;

[0045] Figure 4 This is a schematic diagram of B-spline curve path optimization;

[0046] Figure 5It is a simulation experiment result diagram of the present invention;

[0047] Figure 6 This is a simulation experiment diagram of the robot arm of the present invention operating in a ship cabin. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] In the description of the present invention, it is necessary to understand that the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0050] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0051] like Figure 1 As shown, a method for obstacle avoidance motion path planning of a ship cabin welding robot arm comprises the following steps:

[0052] Step 1: Set the starting point and target point of the two random trees: Determine the starting point x of the random tree T1 according to the starting point of the robot arm start1 and the target point x of random tree T2 goal2 , determine the starting point x of random tree T2 according to the welding end point of the robot arm start2 and the target point x of random tree T1 goal1 ;

[0053] Step 2: Generate random trees with bidirectional growth: two starting points x start1 and x start2 and their corresponding target points xgoal1 and x goal2 Perform bidirectional growth and generate two random trees T respectively 1 and T 2 ;

[0054] Bidirectional growth means exploring and expanding outward from the starting point and the target point respectively until the two trees meet and the planning algorithm converges, which improves the efficiency of finding the path.

[0055] Step 3: Generate sampling point x rand And perform target bias on it: Generate random points in the workspace, and according to the preset probability P, move the random points to the target point x goal1 Or the target point x goal2 Bias, respectively obtain the sampling point x under the preset probability P rand , specifically:

[0056] Since the path search of the RRT* algorithm in a three-dimensional complex environment has the problems of high randomness and long search time, a dual-target bias strategy is adopted as formula (1) for random sampling:

[0057]

[0058] Where rand is a random number between 0 and 1, p is the set target bias threshold, and size(map) is the size of the map;

[0059] When the value of rand is greater than the set target bias threshold, the random point is used as the sampling point x rand ; Otherwise, the target point x goal1 and x goal2 As the sampling points x of the respective random trees rand .

[0060] Step 4: Find the nearest node x near : For any random tree T 1 or T 2 , find the distance from each sampling point x in step three rand The nearest node x near , that is, with the sampling point x rand The node x with the smallest distance near .

[0061] Step 5: Use the improved artificial potential field function to calculate the nearest node x near Calculate the combined force F of its gravitational and repulsive forces t :

[0062] Target point x goal For the nearest node x near Apply attraction F ai , and the sampling point x randAlso for the nearest node x near Apply attraction F aj , at the nearest node x near The surrounding obstacles exert a repulsive force F on it ri , comprehensive attraction F ai and the repulsive force F aj , determine the final direction F of node expansion t :

[0063] The artificial potential field includes the gravitational field function U att As shown in formula (2), the repulsive field function U rep As shown in formula (3), the resultant potential field function is shown in formula (4):

[0064]

[0065] U=U att +U rep (4)

[0066] Among them, K a is the gravitational factor; K r is the repulsion factor; d is the distance between the robot arm and the obstacle; d 0 is the range of the repulsive potential field of the obstacle; P r is the position coordinate of the end of the robot arm; P g is the target point position coordinate; a is the improvement factor, d(P r -P g ) is the relative distance between the end of the robot arm and the target point;

[0067] Each nearest node x near The combined force of attraction and repulsion is F t For: F t =F aj +F ri .

[0068] Step 6: According to the calculated resultant force F t Direction, from the nearest node x near Expand in this direction with a fixed step size s to generate a new node x new ;

[0069] Step 7: Generate new node x new Perform collision detection, that is, determine the new node x new Check whether the connection path with its parent node interferes with obstacles. If a collision occurs, abandon the current expansion and return to step 3 to resample and expand. If no collision occurs, optimize the new node x. new and the nearest node x near The path length between them goes to step eight;

[0070] In step seven, the new node x is optimized new and the nearest node x near The path length between is:

[0071] The new node x new Take any nearest node x near Join as parent node with the nearest node x near The corresponding random tree T 1 or T 2 In the new node x new Search for random tree T within the radius r of the center 1 or T 2 In the node, select the node that can make the start1 and x start2 Go to the new node x respectively new The node x with the shortest path min , and the new node x new The parent node is updated to x min , to optimize the path.

[0072] Step 8. Set the maximum number of iterations n max , repeat steps 3 to 7, and check in each iteration whether the two random trees T1 and T2 are successfully grown and the path connection between them is completed. If the maximum number of iterations n max Inside, two random trees T 1 and T 2 The paths are successfully connected, that is, the two random trees start from their respective starting points x start1 and x start2 Towards their respective target points x goa l1 and x goal2 When they meet and connect during the growth process, a preliminary path is generated and sampling stops; if the maximum number of iterations n is not reached, max Inner join the two trees and return to step three.

[0073] like Figure 2 As shown in the figure, it is the generation process of two random trees of the bidirectional RRT* algorithm, as Figure 3 The following is a preliminary path optimization process of this embodiment: Assume that there are 5 nodes: starting point x start1 、Intermediate node x middle1 、Intermediate node x middle2 、Intermediate node x middle3 、Target point x goal1 ,

[0074] It also includes the optimization of the initial path length and the number of turns, specifically:

[0075] Traverse the initial path in step eight except the starting point xstart1 、Target point x goal1 Or starting point x start2 、Target point x goal2 Remove some intermediate nodes, check whether the preliminary path after removing the intermediate nodes has no collision with obstacles, if no collision, update the preliminary path, if collision occurs, restore the intermediate nodes, select other intermediate nodes to remove, until the preliminary path has no collision with obstacles, and complete the optimization of the preliminary path length and number of turns. That is: directly connect x middle1 With x middle3 , if the path is still collision-free, remove x middle2 Update the path; if a collision occurs, remove other intermediate nodes until the preliminary path has no collision with obstacles.

[0076] It also includes optimization of the linearity of the preliminary path, specifically:

[0077] Except for the starting point x start1 、Target point x goal1 Or starting point x start2 、Target point x goal2 , and check whether the path has no collision with obstacles. If there is no collision, remove the nodes between the two non-adjacent intermediate nodes and update the preliminary path. If a collision occurs, cancel the path connection of the two non-adjacent intermediate nodes and reselect non-adjacent intermediate nodes for path connection until the connection path does not collide with obstacles, completing the optimization of the linearity of the preliminary path. That is, for x in the path, middle1 With x middle4 Connect if x middle1 to x middle4 If the path does not collide with the obstacle, delete x middle2 、x middle3 If a collision occurs, the path connection between the two non-adjacent intermediate nodes is cancelled, and non-adjacent intermediate nodes are reselected for path connection until the connection path does not collide with the obstacle.

[0078] It also includes optimization of the initial path smoothness, specifically:

[0079] Preliminary path smoothness pre-optimization:

[0080] From the target point x of the preliminary path goal1 Start with x start1 The intermediate nodes other than the target point x are connected one by one. goal1 If the path connecting to the intermediate node does not collide with the obstacle, remove the target point x. goal1 and all nodes between the intermediate node, and update the path. If the target point x goal1If the target point x collides with the connection path of the intermediate node, goal1 Connect with the previous intermediate node that does not collide, and repeat the above connection steps with the intermediate node that collides as the new target point until the connection path does not collide with the obstacle, completing the optimization of the initial smoothness of the path. This optimization method is the same as starting from the target point x of the initial path. goal2 Start with x start2 The initial path optimization method for connecting the intermediate nodes other than x one by one is the same; that is: goal1 With x middle2 If there is no collision with obstacles on the connecting route between the two points, continue to the previous node x middle1 Initiate a connection. If there is a collision with an obstacle, middle3 Connect and convert x middle2 Repeat the above steps as the new target point until the connecting path does not collide with obstacles.

[0081] like Figure 4 As shown, the B-spline curve is then used to smoothly fit the path after the preliminary path smoothness is pre-optimized, specifically:

[0082]

[0083] Where P i is the control point of the curve, F i,k (t) is the k-order B-spline basis function.

[0084] like Figure 5 and Figure 6 As shown in the algorithm simulation experiment result diagram, according to the obstacle avoidance planning method, the end of the welding robot arm avoided the spherical and square obstacles and successfully reached the target point, and the selected path was optimal.

[0085] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0086] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for obstacle avoidance motion path planning of a ship cabin welding robot arm, characterized in that: The following steps are involved: Step 1: Set the starting point and target point of the two random trees: Determine the starting point x of the random tree T1 according to the starting point of the robot arm start1 and the target point x of random tree T2 goal2 , determine the starting point x of random tree T2 according to the welding end point of the robot arm start2 and the target point x of random tree T1 goal1 ; Step 2: Generate random trees with bidirectional growth: two starting points x start1 and x start2 and their corresponding target points x goal1 and x goal2 Perform bidirectional growth to generate two random trees T1 and T2 respectively; Step 3: Generate sampling point x rand And perform target bias on it: Generate random points in the workspace, and according to the preset probability P, move the random points to the target point x goal1 Or the target point x goal2 Bias, respectively obtain the sampling point x under the preset probability P rand ; Step 4: Find the nearest node x near : For any random tree T1 or T2, find the distance from each sampling point x in step 3 rand The nearest node x near , that is, with the sampling point x rand The node x with the smallest distance near ; Step 5: Use the improved artificial potential field function to calculate the nearest node x near Calculate the combined force F of its gravitational and repulsive forces t ; Step 6: According to the calculated resultant force F t Direction, from the nearest node x near Expand in this direction with a fixed step size s to generate a new node x new ; Step 7: Generate new node x new Perform collision detection, that is, determine the new node x new Check whether the connection path with its parent node interferes with obstacles. If a collision occurs, abandon the current expansion and return to step 3 to resample and expand. If no collision occurs, optimize the new node x. new and the nearest node x near The path length between them goes to step eight; Step 8. Set the maximum number of iterations n max , repeat steps 3 to 7, and check in each iteration whether the two random trees T1 and T2 are successfully grown and the path connection between them is completed. If the maximum number of iterations n max Within, the paths of the two random trees T1 and T2 are successfully connected, that is, the two random trees start from their respective starting points x start1 and x start2 Towards their respective target points x goal1 and x goal2 When they meet and connect during the growth process, a preliminary path is generated and sampling stops; if the maximum number of iterations n is not reached max Inner join the two trees and return to step 3.

2. The obstacle avoidance motion path planning method for a ship cabin welding manipulator according to claim 1 is characterized in that: The step three is specifically as follows: Since the path search of the RRT* algorithm in a three-dimensional complex environment has the problems of high randomness and long search time, a dual-target bias strategy is adopted as formula (1) for random sampling: Where rand is a random number between 0 and l, p is the set target bias threshold, and size(map) is the size of the map. When the value of rand is greater than the set target bias threshold, the random sampling point is used as the sampling point x. rand ;otherwise , the target point x goal1 and x goal2 As the sampling points x of the respective random trees rand .

3. The obstacle avoidance motion path planning method for a ship cabin welding manipulator according to claim 1 is characterized in that: The step five is specifically as follows: Target point x goal For the nearest node x near Apply attraction F ai , and the sampling point x rand Also for the nearest node x near Apply attraction F aj , at the nearest node x near The surrounding obstacles exert a repulsive force F on it ri , comprehensive attraction F ai and the repulsive force F aj , determine the final direction of node expansion F t : The artificial potential field includes the gravitational field function U att As shown in formula (2), the repulsive field function U rep As shown in formula (3), the resultant potential field function is shown in formula (4): U=U att +U rep (4) Among them, K a is the gravitational factor; K r is the repulsive factor; d is the distance between the robot and the obstacle; d0 is the range of the repulsive potential field of the obstacle; P r is the position coordinate of the end of the robot arm; P g is the target point position coordinate; a is the improvement factor, d(P r -P g ) is the relative distance between the end of the robot arm and the target point; Each nearest node x near The combined force of attraction and repulsion is F t For: F t =F aj +F ri .

4. The obstacle avoidance motion path planning method for a ship cabin welding manipulator according to claim 1 is characterized in that: In step seven, the new node x is optimized new and the nearest node x near The path length between is: The new node x new Take any nearest node x near Join as parent node with the nearest node x near In the corresponding random tree T1 or T2; then with the new node x new Search for nodes in random trees T1 or T2 within a radius r, and select the node that can make the node from the starting point x start1 and x start2 To the new node x new The node x with the shortest path min , and the new node x new The parent node is updated to x min , to optimize the path.

5. The obstacle avoidance motion path planning method for a ship cabin welding robot according to claim 1, characterized in that: It also includes the optimization of the initial path length and the number of turns, specifically: Traverse the initial path in step eight except the starting point x start1 、Target point x goal1 Remove some intermediate nodes, check whether the preliminary path after removing the intermediate nodes has no collision with obstacles, if no collision, update the preliminary path, if collision occurs, restore the intermediate nodes, select other intermediate nodes to remove, until the preliminary path has no collision with obstacles, complete the optimization of the preliminary path length and the number of turns, and another starting point x start2 、Target point x goal2 The same optimization method is used for the preliminary path.

6. The obstacle avoidance motion path planning method for a ship cabin welding robot according to claim 1, characterized in that: It also includes optimization of the linearity of the preliminary path, specifically: Except for the starting point x start1 、Target point x goal1 The path is connected by two non-adjacent intermediate nodes other than x, and the path is checked to see if there is no collision with the obstacle. If there is no collision, the node between the two non-adjacent intermediate nodes is removed, and the preliminary path is updated. If a collision occurs, the path connection between the two non-adjacent intermediate nodes is cancelled, and non-adjacent intermediate nodes are reselected for path connection until the connection path does not collide with the obstacle, completing the optimization of the linearity of the preliminary path. start2 、Target point x goal2 The same optimization method is used for the preliminary path.

7. The obstacle avoidance motion path planning method for a ship cabin welding robot according to claim 1, characterized in that: It also includes optimization of the initial path smoothness, specifically: Preliminary path smoothness pre-optimization: From the target point x of the preliminary path goal1 Start with x start1 The intermediate nodes other than the target point x are connected one by one. goal1 If the path connecting to the intermediate node does not collide with the obstacle, remove the target point x. goal1 and all nodes between the intermediate node, and update the path. If the target point x goal1 If the target point x collides with the connection path of the intermediate node, goal1 Connect with the previous intermediate node that does not collide, and repeat the above connection steps with the intermediate node that collides as the new target point until the connection path does not collide with the obstacle, completing the optimization of the initial smoothness of the path. This optimization method is the same as starting from the target point x of the initial path. goal2 Start with x start2 The preliminary path optimization method for connecting the intermediate nodes other than the first node one by one is the same; Then, the B-spline curve is used to smoothly fit the path after the preliminary path smoothness is pre-optimized, specifically: Where: P i is the control point of the curve, F i,k (t) is the k-order B-spline basis function.

Images