CN115202365B - Obstacle avoidance and optimal path planning method for robot welding based on constructing three-dimensional discrete points - Google Patents

Abstract

The present invention discloses a robot welding obstacle avoidance optimization path planning method based on constructing three-dimensional discrete points. The first step is to use the grid method to model the three-dimensional path planning space and divide it into obstacle discrete points and free discrete points according to whether it contains obstacles; the second step is to unevenly distribute the information concentration of each discrete point during initialization; the third step is to add a feasibility factor to the heuristic function and add the welding gun self-rotation angle influence factor to the path transfer probability to obtain an improved state transfer probability; the fourth step is to use the improved pheromone update rule to obtain an improved welding path; the fifth step is to use the dynamic welding path turning point evaluation function to screen the improved welding path turning points, take the welding path turning point with the highest evaluation, and obtain the global optimal welding obstacle avoidance path for two adjacent welding points. The present invention solves the problem that the global path planning in the robot welding process cannot effectively avoid dynamic obstacles and the path to avoid obstacles is too long.

Description

Obstacle avoidance and optimal path planning method for robot welding based on constructing three-dimensional discrete points

Technical Field

The invention relates to the technical field of robot welding path planning, in particular to a robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points.

Background technique

When the robot welds the car body, it is necessary to avoid obstacles on the path between two adjacent welding points to prevent the welding gun from colliding with the fixture or parts. There are many existing algorithms for robot welding path planning, such as A* algorithm, artificial potential field algorithm, etc., but most of them are based on two-dimensional environment, and there are many limitations for complex three-dimensional space. With the development of algorithms, ant colony algorithm and genetic algorithm have emerged. Among them, ant colony algorithm has been widely used in robot welding path planning because of its strong adaptability and easy combination with other methods. However, it also has many shortcomings. For example, although the planned path can avoid obstacles, it will increase the number of robot empty paths, resulting in an increase in the overall time of robot welding and the time for engineers to debug equipment. In addition, when performing global path planning during robot welding, there are problems such as inability to effectively avoid dynamic obstacles and too long paths to avoid obstacles.

Summary of the invention

In view of the above-mentioned defects, the present invention proposes a robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points, which aims to solve the problems of being unable to effectively avoid dynamic obstacles and the path for avoiding obstacles being too long during global path planning in the robot welding process.

To achieve this object, the present invention adopts the following technical solutions:

The robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points includes the following steps:

Step S1: The three-dimensional path planning space is modeled using a grid method, and is divided into obstacle discrete points and free discrete points according to whether obstacles are contained;

Step S2: unevenly distribute the information density of each discrete point during initialization;

Step S3: Based on the improved information concentration, the feasibility factor is added to the heuristic function and the welding gun self-rotation angle influence factor is added to the path transfer probability to obtain the improved state transfer probability;

Step S4: using the improved pheromone update rule and combining it with the improved state transition probability to obtain an improved welding path;

Step S5: using a dynamic welding path turning point evaluation function to screen the improved welding path turning points, selecting the welding path turning point with the highest evaluation, obtaining a global optimal welding obstacle avoidance path between two adjacent welding points, and covering all welding points.

Preferably, step S1 specifically includes the following steps:

Step S11: Establish a three-dimensional coordinate system by taking any vertex in the three-dimensional space as the coordinate origin;

Step S12: using a plane perpendicular to the x-axis and a plane perpendicular to the y-axis in the three-dimensional coordinate system to evenly divide the three-dimensional space to obtain a plurality of cubic grids of equal volume, wherein each vertex of the cubic grid is used as a discrete point in welding path planning;

Step S13: Select the discrete points required for welding path planning and determine the type of the selected discrete points; if the workpiece model covers the discrete points of the cubic grid, it indicates that there is an obstacle at the discrete point on the workpiece model, and the discrete point is an obstacle discrete point; if the workpiece model does not cover the discrete points of the cubic grid, it indicates that there is no obstacle at the discrete point on the workpiece model, and the discrete point is a free discrete point.

Preferably, step S2 specifically includes the following steps:

Step S21: connect two adjacent welding points to obtain a line segment L, and take a plane perpendicular to the x-axis to divide the three-dimensional space into equal parts;

Step S22: construct a circle with the intersection of L and each plane as the center and _Lm as the radius, and take the points in the circle as the better discrete points. Different initial pheromone concentrations are divided according to the distance from the better discrete points to the line segment L. The mathematical expression is as follows:

Where, τ _{0 is} the initial pheromone concentration, is the improved initialization pheromone concentration, _Lymax and _Lzmax are the maximum lateral movement distance and maximum longitudinal movement distance of segment L, respectively, dis(H _o ,L _m ) is the distance from discrete point H _o to segment L, and α is the weighting value, which is determined according to the specific parameters of the three-dimensional space.

Preferably, in step S3, a feasibility factor is added to construct a new heuristic function, and the specific mathematical expression is as follows:

Where η _ij is the heuristic function, which represents the expected degree of moving from the current discrete point i to the next discrete point j; d _ij is the distance from the current discrete point i to the next discrete point j; is the magnification; is the feasibility influencing factor, and different magnification factors are taken according to the surrounding discrete points of the next discrete point j.

Preferably, in step S3, adding the welding gun self-rotation angle influence factor to the path transfer probability specifically includes the following steps:

Step S31: Adopting interval Represents the power range near the minimum average power consumption, where is the minimum value of the average power consumption of the robot; is the difference between the maximum and minimum values of average power consumption;

Step S32: Take the welding gun starting point and end point corresponding to the power interval near the minimum value, the welding gun rotation angles at the welding gun starting point and end point are γ _a and γ _b respectively; the specific mathematical formula of the welding gun rotation angle influencing factor is as follows:

Among them, _Vij represents the influencing factor of the welding gun rotation angle; _γi represents the welding gun rotation angle of the current discrete point i; _γj represents the welding gun rotation angle of the next discrete point j.

Preferably, in step S3, according to the new heuristic function with added feasibility factor and the influencing factor of welding gun self-rotation angle, the improved state transition probability formula is as follows:

in, represents the probability of moving from discrete point i to discrete point j at time t; τ _ij (t) represents the information density of moving from discrete point i to discrete point j at time t; _dij represents the distance between discrete point i and discrete point j; _ηij (t) represents the degree of inspiration, and its value is the reciprocal of the distance between discrete point i and discrete point j, that is, _ηij (t) = 1/ _dij ; _Vij (t) represents the influence factor of the welding gun self-rotation angle at time t; α is used to control the information density; β is used to control the path visibility; δ is the importance of the welding gun self-rotation angle influence factor, which is appropriately selected according to the actual welding situation of the robot; _dk is the set of all discrete points that can be directly reached by the kth discrete point in the next step.

Preferably, step S4 specifically includes the following steps:

Arrange all path lengths in ascending order, select the paths in the front part for pheromone update, and the update rules are as follows:

τ _ijg = (1-ρ)τ _ijg +ρΔτ _ijg (7)

Among them, τ _ijg is the pheromone concentration left over from the g-th path after the complete welding path; Δτ _ijg is the pheromone content assigned to the g-th path; len(g) is the path length of the g-th path; rank(g) is the ranking of path g among all paths; Num _r is the number of paths to be updated; ρ is the pheromone volatilization factor, and the initial value of ρ is set to 0.9; Q is the pheromone constant.

Preferably, in step S5, the specific mathematical formula of the dynamic welding path turning point evaluation function is as follows:

H(ψ,ω)=ξ*(θ*Gunangle(ω)+λ*Wpointangle(ψ)) (9)

Among them, H(ψ, ω) represents the dynamic welding path turning point evaluation function; Gunangle(v, ω) is the welding gun opening and closing angle evaluation function, which represents the difference between the welding gun opening and closing angles between the two path turning points; Wpointangle(v, ω) is the path angle evaluation function, which represents the dynamic angle between the direction vector formed by the adjacent path turning points and the two starting and ending welding point vectors; ξ is the smoothing coefficient; θ and λ are the weighted values of each evaluation function, which are weighted according to the actual situation; ω represents the absolute value of the difference between the welding gun opening and closing angles; ψ is the absolute value of the angle between the two vectors; φ is an angle parameter greater than 0; ψ is an angle parameter greater than 0; Construct vectors for two adjacent path turning points, is the fixed vector of the two welding points.

The technical solution provided by the embodiments of the present application may have the following beneficial effects:

This scheme is first based on the improved welding path planning design and the selection of appropriate path turning points, and then further optimizes the selected path turning points based on the path turning point selection evaluation scheme, so as to achieve the optimization of the welding path design under the condition of obstacles, and select the path with the shortest welding path, the lowest average power consumption of the robot, and the smallest opening and closing angle difference of the welding gun.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a step diagram of a robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points;

FIG. 2 is a schematic diagram of an embodiment.

Detailed ways

The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

The robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points includes the following steps:

Step S1: The three-dimensional path planning space is modeled using a grid method, and is divided into obstacle discrete points and free discrete points according to whether obstacles are contained;

Step S2: unevenly distribute the information density of each discrete point during initialization;

Step S3: Based on the improved information concentration, the feasibility factor is added to the heuristic function and the welding gun self-rotation angle influence factor is added to the path transfer probability to obtain the improved state transfer probability;

Step S4: using the improved pheromone update rule and combining it with the improved state transition probability to obtain an improved welding path;

Step S5: using a dynamic welding path turning point evaluation function to screen the improved welding path turning points, selecting the welding path turning point with the highest evaluation, obtaining a global optimal welding obstacle avoidance path between two adjacent welding points, and covering all welding points.

The robot welding obstacle avoidance optimization path planning method based on the construction of three-dimensional discrete points in this scheme, the first step is to use the grid method to model the three-dimensional path planning space, and divide it into obstacle discrete points and free discrete points according to whether it contains obstacles, which provides a basis for selecting the optimal welding obstacle avoidance path and improves the search efficiency of the obstacle avoidance path algorithm. The second step is to distribute the information concentration of each discrete point unevenly during initialization, which can simplify the complexity of the algorithm, improve the efficiency of the early search and accelerate the finding of the general direction of the welding path. The third step is based on the improved information concentration, and add the feasibility factor to the heuristic function and add the welding gun self-rotation angle influence factor to the path transfer probability to obtain the improved state transfer probability, wherein the feasibility factor is added to the heuristic function, and different amplification parameters are given according to the distribution of the obstacle discrete points around the current discrete point, so as to improve the difference in the heuristic information of adjacent discrete points, which is conducive to avoiding the algorithm from falling into the local optimum; the welding self-rotation angle influence factor is added to make the robot consume the least energy and the smallest average power in the obtained welding obstacle avoidance path. The fourth step is to use the improved pheromone update rule and combine it with the improved state transition probability to obtain an improved welding path. The improved welding path is the path with the shortest length and the least number of turns, so as to obtain a path with higher smoothness and the least power consumption of the robot. The fifth step is to use the dynamic welding path turning point evaluation function to screen the improved welding path turning points, select the welding path turning point with the highest evaluation, obtain the global optimal welding obstacle avoidance path of two adjacent welding points, and cover all welding points, so as to ensure the rationality and optimality of the obtained welding obstacle avoidance path.

This scheme is first based on the improved welding path planning design and the selection of appropriate path turning points, and then further optimizes the selected path turning points based on the path turning point selection evaluation scheme, so as to achieve the optimization of the welding path design under the condition of obstacles, and select the path with the shortest welding path, the lowest average power consumption of the robot, and the smallest opening and closing angle difference of the welding gun.

Preferably, step S1 specifically includes the following steps:

Step S11: Establish a three-dimensional coordinate system by taking any vertex in the three-dimensional space as the coordinate origin;

Step S12: evenly divide the three-dimensional space by using a plane perpendicular to the x-axis and a plane perpendicular to the y-axis in the three-dimensional coordinate system to obtain a plurality of cubic grids of equal volume, wherein each vertex of the cubic grid is used as a discrete point in welding path planning;

Step S13: Select the discrete points required for welding path planning and determine the type of the selected discrete points; if the workpiece model covers the discrete points of the cubic grid, it indicates that there is an obstacle at the discrete point on the workpiece model, and the discrete point is an obstacle discrete point; if the workpiece model does not cover the discrete points of the cubic grid, it indicates that there is no obstacle at the discrete point on the workpiece model, and the discrete point is a free discrete point.

In one embodiment, as shown in FIG2 , the lower left vertex of the three-dimensional map is used as the coordinate origin O to establish a three-dimensional coordinate system xyz, and the three-dimensional space is divided into n equal parts by using the plane perpendicular to the x-axis and the plane perpendicular to the y-axis in the three-dimensional coordinate system, and is divided into a number of cubic grids of equal volume. Each vertex A, B, C, etc. of the cubic grid is a discrete point in welding path planning, which replaces the nodes in the traditional welding path planning algorithm and facilitates path planning. The discrete points A, B and C required for welding path planning are selected. If the workpiece model covers the discrete points A, B and C of the cubic grid, it indicates that the discrete points A, B and C have obstacles on the workpiece model, and the discrete points A, B and C are obstacle discrete points; if the workpiece model does not cover the discrete points A, B and C of the cubic grid, it indicates that the discrete points A, B and C do not have obstacles on the workpiece model, and the discrete points A, B and C are free discrete points.

Preferably, step S2 specifically includes the following steps:

Step S21: connect two adjacent welding points to obtain a line segment L, and take a plane perpendicular to the x-axis to divide the three-dimensional space into equal parts;

Step S22: construct a circle with the intersection of L and each plane as the center and _Lm as the radius, and take the points in the circle as the better discrete points. Different initial pheromone concentrations are divided according to the distance from the better discrete points to the line segment L. The mathematical expression is as follows:

Where, τ _{0 is} the initial pheromone concentration, is the initialization pheromone concentration after improvement, _Lymax and _Lzmax are the maximum lateral moving distance and maximum longitudinal moving distance of line segment L, dis(H _o ,L _m) is the distance from discrete point H _o to line segment L, and α is the weighting value, which is determined according to the specific parameters of the three-dimensional space.

In the traditional welding path planning algorithm, the initialization pheromone concentration is the same, which makes the walking randomness too strong and the convergence speed too slow during the search process. In addition, the pheromone carrier is a line segment between discrete points, which greatly increases the spatial complexity of the algorithm. The improved algorithm in this scheme stores pheromones in discrete points and distributes the pheromones of each point unevenly during initialization, thereby simplifying the algorithm complexity, improving the early search efficiency and accelerating the finding of the general direction of the welding path.

Preferably, in step S3, a feasibility factor is added to construct a new heuristic function, and the specific mathematical expression is as follows:

Where η _ij is the heuristic function, which represents the expected degree of moving from the current discrete point i to the next discrete point j; d _ij is the distance from the current discrete point i to the next discrete point j; is the magnification; is the feasibility influencing factor, and different magnification factors are taken according to the surrounding discrete points of the next discrete point j.

The heuristic function is an important part of the welding path planning algorithm. Its role is to use distance information to guide and select the shortest path, which directly affects the convergence, stability and optimality of the algorithm. The heuristic function η _ij of the traditional welding path planning algorithm is inversely proportional to the distance between discrete points i and j. However, when the distance difference between adjacent discrete points is very small, it has little effect on guiding the path. At the same time, when searching for discrete points, it often ignores the surrounding obstacles and gives priority to the discrete points closest to the current discrete point, thus falling into the local optimum. To address this problem, this scheme adds feasibility factors to construct a new heuristic function, increases the difference in heuristic information between adjacent discrete points, and makes the search for the optimal path more directional.

Preferably, in step S3, adding the welding gun self-rotation angle influence factor to the path transfer probability specifically includes the following steps:

Step S31: Adopting interval Represents the power range near the minimum average power consumption, where is the minimum value of the robot’s average power consumption; is the difference between the maximum and minimum values of average power consumption;

Step S32: Take the welding gun starting point and end point corresponding to the power interval near the minimum value, the welding gun rotation angles at the welding gun starting point and end point are γ _a and γ _b respectively; the specific mathematical formula of the welding gun rotation angle influencing factor is as follows:

Among them, _Vij represents the influencing factor of the welding gun rotation angle; _γi represents the welding gun rotation angle of the current discrete point i; _γj represents the welding gun rotation angle of the next discrete point j.

Specifically, the welding wire rotates around the axis during arc welding, and the angle of rotation is called the welding gun self-rotation angle. This angle does not affect the quality of the welding process, but it makes a difference to the energy consumption of the robot during welding, mainly because when the welding gun self-rotation angle changes, the corresponding positions, rotation angles and speeds of the robot's joint axes are different. Randomly selecting a set of welding gun self-rotation angles at the starting and ending points of the welding gun will cause the robot to have different average power during welding. In actual production, it is expected to obtain a combination of starting and ending welding gun self-rotation angles with the minimum average power consumption. In this scheme, a welding self-rotation angle influence factor is added. The closer the value of the welding gun self-rotation angle influence factor is to 1, the less power the robot consumes when welding through these two discrete points, so that the robot consumes the least energy and the lowest average power in the obtained welding obstacle avoidance path.

Preferably, in step S3, according to the new heuristic function with added feasibility factor and the influencing factor of welding gun self-rotation angle, the improved state transition probability formula is as follows:

in, represents the probability of moving from discrete point i to discrete point j at time t; τ _ij (t) represents the information density of moving from discrete point i to discrete point j at time t; _dij represents the distance between discrete point i and discrete point j; _ηij (t) represents the degree of inspiration, and its value is the reciprocal of the distance between discrete point i and discrete point j, that is, _ηij (t) = 1/ _dij ; _Vij (t) represents the influence factor of the welding gun self-rotation angle at time t; α is used to control the information density; β is used to control the path visibility; δ is the importance of the welding gun self-rotation angle influence factor, which is appropriately selected according to the actual welding situation of the robot; _dk is the set of all discrete points that can be directly reached by the kth discrete point in the next step.

In this scheme, the feasibility factor is introduced into the heuristic function, and the influencing factor of the welding self-rotation angle is added to improve the state transition probability function, thereby increasing the possibility of finding the optimal path.

Preferably, step S4 specifically includes the following steps:

Arrange all path lengths in ascending order, select the paths in the front part for pheromone update, and the update rules are as follows:

τ _ijg = (1-ρ)τ _ijg +ρΔτ _ijg (7)

Among them, τ _ijg is the pheromone concentration left over from the g-th path after the complete welding path; Δτ _ijg is the pheromone content assigned to the g-th path; len(g) is the path length of the g-th path; rank(g) is the ranking of path g among all paths; Num _r is the number of paths to be updated; ρ is the pheromone volatilization factor, and the initial value of ρ is set to 0.9; Q is the pheromone constant.

When a path search is completed, all path lengths are sorted in ascending order using the length as the evaluation value, and only the paths in the front are selected for pheromone update. Specifically, the total number of iterations required G (0＜g＜G) is first set. After each iteration, the value of g will increase by one. After the iteration, all paths will be sorted. If the welding path lengths of two paths are similar, they will be sorted according to the number of turns. Paths with many turns will increase the power consumption of the robot. Paths with fewer turns should be selected in front. Finally, the path with the shortest length and the least number of turns will be selected as the optimal path, so as to obtain a path with higher smoothness and the least power consumption of the robot.

Preferably, in step S5, the specific mathematical formula of the dynamic welding path turning point evaluation function is as follows:

H(ψ,ω)=ξ*(θ*Gunangle(ω)+λ*Wpointangle(ψ)) (9)

Among them, H(ψ, ω) represents the dynamic welding path turning point evaluation function; Gunangle(v, ω) is the welding gun opening and closing angle evaluation function, which represents the difference between the welding gun opening and closing angles between the two path turning points; Wpointangle(v, ω) is the path angle evaluation function, which represents the dynamic angle between the direction vector formed by the adjacent path turning points and the two starting and ending welding point vectors; ξ is the smoothing coefficient; θ and λ are the weighted values of each evaluation function, which are weighted according to the actual situation; ω represents the absolute value of the difference between the welding gun opening and closing angles; ψ is the absolute value of the angle between the two vectors; φ is an angle parameter greater than 0; ψ is an angle parameter greater than 0; Construct vectors for two adjacent path turning points, is the fixed vector of the two welding points.

The design principle of the evaluation function is that the welding gun at the end of the robot can avoid obstacles such as fixtures and workpieces as much as possible, and move quickly toward the next welding point. During the welding process, the difference between the opening and closing angles of the welding gun should not be too large, and the turning angle of each path turning point should not be too large. Therefore, in this embodiment, the values of θ and λ are 1. If the deviation of the opening and closing angles of the welding gun is too large, θ can be appropriately reduced; in the early stage of path planning, the distance to the next welding point is far, and λ should take a smaller value; in the later stage of path planning, the distance to the next welding point is close, and λ should take a larger value. Further explanation, when the evaluation function of a path turning point is too low, the point can be discarded, and the path turning point can be re-searched with the previous path turning point as the starting point, and the path turning point can be reversely searched with the next path turning point as the end point. The two are combined to deduce a suitable intermediate turning point.

In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims (4)

1. A robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points, characterized in that it includes the following steps: Step S1: The three-dimensional path planning space is modeled using a grid method, and is divided into obstacle discrete points and free discrete points according to whether obstacles are contained; Step S2: unevenly distribute the pheromone concentration of each discrete point during initialization; Step S3: Based on the improved pheromone concentration, the feasibility factor is added to the heuristic function and the welding gun self-rotation angle influence factor is added to the path transfer probability to obtain the improved state transfer probability; Step S4: using the improved pheromone update rule and combining it with the improved state transition probability to obtain an improved welding path; Step S5: using a dynamic welding path turning point evaluation function to screen the improved welding path turning points, selecting the welding path turning point with the highest evaluation, obtaining a global optimal welding obstacle avoidance path between two adjacent welding points, and covering all welding points; In step S3, a feasibility factor is added to construct a new heuristic function. The specific mathematical expression is as follows: Where η _ij is the heuristic function, which represents the expected degree of moving from the current discrete point i to the next discrete point j; d _ij is the distance from the current discrete point i to the next discrete point j; is the magnification; is the feasibility influencing factor, and takes different magnifications according to the surrounding discrete points of the next discrete point j; In step S3, the influence factor of the welding gun self-rotation angle is added to the path transfer probability, which specifically includes the following steps: Step S31: Adopting interval Represents the power range near the minimum average power consumption, where is the minimum value of the average power consumption of the robot; is the difference between the maximum and minimum values of average power consumption; Step S32: Take the welding gun starting point and end point corresponding to the power interval near the minimum value, the welding gun rotation angles at the welding gun starting point and end point are γ _a and γ _b respectively; the specific mathematical formula of the welding gun rotation angle influencing factor is as follows: Among them, _Vij represents the influencing factor of the welding gun rotation angle; _γi represents the welding gun rotation angle of the current discrete point i; _γj represents the welding gun rotation angle of the next discrete point j; In step S3, according to the new heuristic function with added feasibility factor and the influencing factor of welding gun self-rotation angle, the improved state transition probability formula is as follows: in, represents the probability of moving from discrete point i to discrete point j at time t; τ _ij (t) represents the pheromone concentration moving from discrete point i to discrete point j at time t; _dij represents the distance between discrete point i and discrete point j; _ηij (t) represents the degree of inspiration, and its value is the inverse of the distance between discrete point i and discrete point j, that is, _ηij (t) = 1/ _dij ; _Vij (t) represents the influence factor of the welding gun self-rotation angle at time t; α is used to control the pheromone concentration; β is used to control the path visibility; δ is the importance of the influence factor of the welding gun self-rotation angle, which is appropriately selected according to the actual welding situation of the robot; _dk is the set of all discrete points that can be directly reached by the kth discrete point in the next step; Step S4 specifically includes the following steps: Arrange all path lengths in ascending order, select the paths in the front part for pheromone update, and the update rules are as follows: τ _ijg = (1-ρ)τ _ijg +ρΔτ _ijg (7) Among them, τ _ijg is the pheromone concentration left over from the g-th path after the complete welding path; Δτ _ijg is the pheromone content assigned to the g-th path; len(g) is the path length of the g-th path; rank(g) is the ranking of path g among all paths; Num _r is the number of paths to be updated; ρ is the pheromone volatilization factor, and the initial value of ρ is set to 0.9; Q is the pheromone constant. 2. The robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points according to claim 1 is characterized in that: in step S1, the following steps are specifically included: Step S11: Establish a three-dimensional coordinate system by taking any vertex in the three-dimensional space as the coordinate origin; Step S12: using a plane perpendicular to the x-axis and a plane perpendicular to the y-axis in the three-dimensional coordinate system to evenly divide the three-dimensional space to obtain a plurality of cubic grids of equal volume, wherein each vertex of the cubic grid is used as a discrete point in welding path planning; Step S13: Select the discrete points required for welding path planning and determine the type of the selected discrete points; if the workpiece model covers the discrete points of the cubic grid, it indicates that there is an obstacle at the discrete point on the workpiece model, and the discrete point is an obstacle discrete point; if the workpiece model does not cover the discrete points of the cubic grid, it indicates that there is no obstacle at the discrete point on the workpiece model, and the discrete point is a free discrete point. 3. The robot welding obstacle avoidance and optimal path planning method based on constructing three-dimensional discrete points according to claim 1 is characterized in that: in step S2, the following steps are specifically included: Step S21: connect two adjacent welding points to obtain a line segment L, and take a plane perpendicular to the x-axis to divide the three-dimensional space into equal parts; Step S22: construct a circle with the intersection of L and each plane as the center and _Lm as the radius, and take the points in the circle as the better discrete points. Different initial pheromone concentrations are divided according to the distance from the better discrete points to the line segment L. The mathematical expression is as follows: Where, τ _{0 is} the initial pheromone concentration, is the improved initialization pheromone concentration, _Lymax and _Lzmax are the maximum lateral movement distance and maximum longitudinal movement distance of line segment L, respectively, dis(H _o , L _m ) is the distance from discrete point H _o to line segment L, and α is the weighting value, which is determined according to the specific parameters of the three-dimensional space. 4. The robot welding obstacle avoidance and optimization path planning method based on constructing three-dimensional discrete points according to claim 1 is characterized in that: in step S5, the specific mathematical formula of the dynamic welding path turning point evaluation function is as follows: H(ψ,ω)=ξ*(θ*Gunangle(ω)+λ*Wpointangle(ψ)) (9) Among them, H(ψ, ω) represents the dynamic welding path turning point evaluation function; Gunangle(ω) is the welding gun opening and closing angle evaluation function, which represents the difference between the welding gun opening and closing angles between the two path turning points; Wpointangle(ψ) is the path angle evaluation function, which represents the dynamic angle between the direction vector formed by the adjacent path turning points and the two starting and ending welding point vectors; ξ is the smoothing coefficient; θ and λ are the weighted values of each evaluation function, which are weighted according to the actual situation; ω represents the absolute value of the difference between the welding gun opening and closing angles; ψ is the absolute value of the angle between the two vectors; is an angle parameter greater than 0; Construct vectors for two adjacent path turning points, is the fixed vector of the two welding points.