Multi-robot collaborative handling method and system for irregular objects based on multi-objective optimization

Abstract

The present invention discloses a multi-robot collaborative handling method and system for irregular objects based on multi-objective optimization, belonging to the technical field of multi-mobile robot collaborative control, including: obtaining the shape information of an object and extracting the edge contour coordinates of the object; converting the edge contour coordinates of the object into parametric equations of the actual action points at the ends of the robot links; expanding the object shape, establishing a mapping relationship between the parametric equations and reachable positions, and obtaining a set of actual reachable position points of the robot; constructing a multi-objective optimization function, using the coordinate points as variables to be optimized, optimizing with an improved NSGA-II algorithm to obtain the optimal solution, and determining the best action point coordinates according to the number of robots; after obtaining the best action point coordinates, controlling the multi-robots to perform rigid formation to complete the collaborative handling task of irregular objects. The present invention solves the problems of missing object shape analysis, insufficient formation constraints, and multi-objective collaborative optimization in the prior art, thereby realizing the efficiency and stability of the handling task.

Description

Technical Field

[0001] The present invention belongs to the technical field of cooperative control of multi-mobile robots, and particularly relates to a method and system for multi-robot cooperative handling of irregular objects based on multi-objective optimization. Background Technique

[0002] The statements here only provide the background technology related to the present invention and do not necessarily constitute the prior art.

[0003] In recent years, with the continuous development of robot technology, multi-robot systems have become a research hotspot in the global fields of robotics and control. Cooperative handling is one of the popular research applications of multi-robot systems, and its goal is to move a target object to a specific destination through a group of robots. Its advantage lies in being able to handle objects that are too heavy for a single robot to carry independently. Designing and implementing a cooperative handling system requires considering many factors, such as selecting a multi-robot communication mechanism, coordination algorithm, control architecture, task allocation mechanism, handling platform, etc.

[0004] The multi-robot cooperative handling technology has become a research hotspot due to its scalability and high efficiency. The existing mainstream handling strategies include the "only push" strategy (pushing the object by the robot body), the "caging" strategy (surrounding the object at the center of the formation), and the "grasping" strategy (using a high-degree-of-freedom robotic arm to grasp the object). However, the inventor found that these strategies have the following defects: the lack of object shape modeling, the lack of precise modeling of the geometric features of irregular objects, resulting in the lack of a theoretical basis for the selection of the action point; insufficient rigid constraints, the robot formation does not consider the adaptability to the object shape, easily leading to insufficient stability or low efficiency during the handling process; the blank of multi-objective cooperative optimization, the existing methods do not comprehensively weigh key indicators such as system energy consumption, handling efficiency, and safety. Summary of the Invention

[0005] The purpose of the present invention is to overcome the above-mentioned deficiencies existing in the prior art, and provide a method and system for multi-robot cooperative handling of irregular objects based on multi-objective optimization. By accurately modeling the object shape, optimizing the action positions of multiple robots, constructing a multi-objective optimization function and using an improved NSGA-II algorithm to obtain the optimal solution, and designing a DMPC controller with rigid formation constraints to complete the cooperative handling task, it solves the problems of missing object shape analysis, insufficient formation constraints, and multi-objective cooperative optimization in the prior art, thereby realizing the high efficiency and stability of the handling task.

[0006] To achieve the above purpose, the present invention is realized through the following technical solutions:

[0007] On the one hand, the technical solution of the present invention provides a method for multi-robot cooperative handling of irregular objects based on multi-objective optimization, including:

[0008] Obtain the shape information of the object and extract the edge contour coordinates of the object;

[0009] Convert the edge contour coordinates of the object into the parametric equations of the actual action points at the end of the robot link;

[0010] Expand the object shape, establish the mapping relationship between the parametric equations and the reachable positions, and obtain the set of actual reachable position points of the robot;

[0011] Construct a multi-objective optimization function, use the coordinate points as the variables to be optimized, optimize the quality of the initial population using the improved NSGA-II algorithm and define the degree of constraint violation, and then obtain the optimal solution, and determine the best action point coordinates according to the number of robots;

[0012] After obtaining the best action point coordinates, control the multi-robots to perform rigid formation to complete the collaborative handling task of irregular objects.

[0013] In at least one embodiment, the obtaining the shape information of the object and extracting the edge contour coordinates of the object is specifically: obtaining the original image of the object through an RGB camera, converting the original image into a grayscale image and performing Gaussian filtering processing, and then extracting the edge contour coordinates of the object through an edge detection algorithm.

[0014] In at least one embodiment, the converting the edge contour coordinates of the object into the parametric equations of the actual action points at the end of the robot link is specifically: according to the object type, adopt the corresponding interpolation method to convert the edge coordinates of the object into the parametric equations of the actual action points at the end of the robot link.

[0015] In at least one embodiment, the object types include closed polygons and smooth closed curves.

[0016] In at least one embodiment, if the object type is a closed polygon, use the piecewise linear interpolation method to construct the parametric equations; if the object type is a smooth closed curve, use the cubic spline interpolation method to construct the parametric equations.

[0017] In at least one embodiment, expanding the object shape, establishing the mapping relationship between the parametric equations and the reachable positions, and obtaining the set of actual reachable position points of the robot is specifically: expanding the object shape described by the parametric equations outward along the normal direction, with the expansion length being the length of the robot link, establishing the mapping relationship between the parametric equations and the reachable positions, and obtaining the candidate two-dimensional coordinate points that the robot can directly reach; eliminating the invalid points caused by the non-convexity of the object or volume conflict, and finally obtaining the set of actual reachable position points after elimination.

[0018] In at least one embodiment, the constructing the multi-objective optimization function is specifically:

[0019] Construct a multi-objective optimization function based on the corresponding positions of each robot and the position corresponding to the centroid of the object; maximize the distance between robots to establish a position dispersion objective function; based on making each robot as far away as possible from the centroid of the irregular object, establish a torque maximization objective function; based on the symmetric distribution of the acting forces, establish a shape closure objective function; add collision constraints between the robots to ensure that the distance between the robots meets the safety distance threshold.

[0020] In at least one embodiment, based on the constructed multi-objective function and collision constraints, use the improved NSGA-II algorithm to take the object parametric equation and the number of robots as inputs. First, iterate to obtain the optimal solutions of each objective function as individuals in the initial population, then iterate to solve for the Pareto front, and according to the actual task requirements, adjust the weights of the position dispersion objective function, torque maximization objective function, and shape closure objective function on the Pareto front, and determine the optimal solution through weighted summation to obtain the best reachable position coordinates and best action point coordinate information of the multi-robots.

[0021] In at least one embodiment, after obtaining the best action point coordinates, rigidly connect the end of the robot link to the object, and use a controller with a rigid formation hard constraint to adjust the formation attitude in real time to control the multi-robots to complete the collaborative handling task of the irregular object.

[0022] On the other hand, the technical solution of the present invention also provides a multi-robot collaborative handling system for irregular objects based on multi-objective optimization, including:

[0023] An edge contour coordinate extraction module, configured to: obtain the shape information of the object and extract the edge contour coordinates of the object;

[0024] A parametric equation construction module, configured to: convert the edge contour coordinates of the object into a parametric equation of the actual action point at the end of the robot link;

[0025] A reachable position point set generation module, configured to: expand the shape of the object, establish a mapping relationship between the parametric equation and the reachable position, and obtain the set of actual reachable position points of the robot;

[0026] A multi-objective optimization module, configured to: construct a multi-objective optimization function, use the coordinate points as variables to be optimized, use the improved NSGA-II algorithm to optimize the quality of the initial population and define the constraint violation degree, and then obtain the optimal solution, and determine the best action point coordinates according to the number of robots;

[0027] A task execution module, configured to: after obtaining the best action point coordinates, control the multi-robots to form a rigid formation to complete the collaborative handling task of the irregular object.

[0028] The beneficial effects of the above technical solution of the present invention are as follows:

[0029] 1) The multi-robot collaborative handling method and system for irregular objects based on multi-objective optimization of the present invention accurately models the object shape, optimizes the action positions of multiple robots, constructs a multi-objective optimization function, and uses an improved NSGA-II algorithm to obtain the optimal solution. It also designs a DMPC controller with rigid formation constraints to complete the collaborative handling task, solving the problems of missing object shape analysis, insufficient formation constraints, and multi-objective collaborative optimization in the prior art, thereby achieving the efficiency and stability of the handling task.

[0030] 2) The multi-robot collaborative handling method and system for irregular objects based on multi-objective optimization of the present invention can make the shape modeling efficient. By compressing and storing the contour data of irregular objects using parametric equations, the computational complexity is reduced. Using the dilation of the irregular object shape, the relationship between the solution of the optimization problem and the actual positions of the mobile robots and the action point positions is clearly established, facilitating the subsequent expansion of the position allocation and trajectory control modules.

[0031] 3) The multi-robot collaborative handling method and system for irregular objects based on multi-objective optimization of the present invention can achieve multi-objective collaborative optimization, comprehensively balance the stability, efficiency, and safety of the overall collaborative handling system, and improve the system adaptability.

[0032] 4) The multi-robot collaborative handling method and system for irregular objects based on multi-objective optimization of the present invention can enhance the practicality of the optimization algorithm and significantly shorten the solution time. In actual industrial scenarios, it can be applied to the collaborative handling of large irregular objects by multiple AGVs, having wide engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The specification drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0034] Figure 1 FIG. is the overall flowchart of the multi-robot collaborative handling method for irregular objects based on multi-objective optimization of the present invention.

[0035] Figure 2 FIG. is the schematic diagram of object contour extraction and action position optimization of the present invention; wherein, (a) is the edge contour diagram of a rectangle-like object and the schematic diagram of the selected action points, (b) is the edge contour diagram of an L-shaped object and the schematic diagram of the selected action points, and (c) is the edge contour diagram of an ellipse-like object and the schematic diagram of the selected action points. DETAILED DESCRIPTION OF THE INVENTION

[0036] It should be noted that the following detailed description is illustrative and aims to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0037] As introduced in the background art, the purpose of the present invention is to overcome the deficiencies existing in the above-mentioned prior art, and to provide a multi-robot collaborative handling method and system for non-regular objects based on multi-objective optimization. By accurately modeling the object shape, optimizing the action positions of multiple robots, constructing a multi-objective optimization function and using an improved NSGA-II algorithm to obtain the optimal solution, and designing a DMPC controller with rigid formation constraints to complete the collaborative handling task, the problems of missing object shape analysis, insufficient formation constraints and multi-objective collaborative optimization in the prior art are solved, thereby realizing the efficiency and stability of the handling task.

[0038] Embodiment 1

[0039] In a typical embodiment of the present invention, as Figure 1 shown, this embodiment discloses a multi-robot collaborative handling method for non-regular objects based on multi-objective optimization, including:

[0040] S100: Obtain the shape information of the object and extract the edge contour coordinates of the object;

[0041] S200: Convert the edge contour coordinates of the object into the parametric equations of the actual action points at the ends of the robot links;

[0042] S300: Expand the object shape, generate a set of reachable position points of the robot, and establish the mapping relationship between the parametric equations and the reachable positions;

[0043] S400: Construct a multi-objective optimization function, use the coordinate points as the variables to be optimized, optimize the quality of the initial population using an improved NSGA-II algorithm and define the constraint violation degree, and then obtain the optimal solution, and determine the best action point coordinates according to the number of robots;

[0044] S500: After obtaining the best action point coordinates, control multiple robots to form a rigid formation to complete the collaborative handling task of non-regular objects.

[0045] The above multi-robot collaborative handling method for non-regular objects based on multi-objective optimization will be described in detail below with specific embodiments.

[0046] S100: Obtain the shape information of the object and extract the edge contour coordinates of the object.

[0047] S101. Image acquisition: Set up an RGB industrial camera with a resolution of 1920*1080 to obtain a top-down image of the irregular object, which is used as the original image of the irregular object.

[0048] S102. Grayscale and denoising of the image: Perform grayscale processing on the original image to obtain the grayscale image of the irregular object, and perform Gaussian filtering on the grayscale image to eliminate the interference of noise on edge detection. Specifically, the conversion method of grayscale is as follows:

[0049] 。

[0050] S103. Extract edge contour coordinates: Use the Canny edge detection algorithm to extract the edge contour coordinates of the irregular object from the denoised grayscale image through steps such as image gradient calculation, non-maximum suppression, and double-threshold edge determination.

[0051] Specifically, the edge contour mainly uses the feature of the drastic change in image grayscale as the basis. The main steps of the Canny algorithm are as follows:

[0052] (1) Calculate the image gradient, and use the Sobel operator to calculate the gradient and direction of each pixel;

[0053] (2) Suppress non-maxima, calculate whether the gradient amplitude of the current pixel point along the gradient direction is a local maximum, and only retain the pixel points with the local maximum gradient value as edges, suppressing other points;

[0054] (3) Determine edges with double thresholds, and filter out non-edge points generated by the image itself or noise through high and low two thresholds.

[0055] S200: Convert the edge contour coordinates of the object into the parametric equation of the actual action point at the end of the robot link.

[0056] Based on the shape information of the irregular object obtained in S100 and the extracted edge contour coordinates, the types of irregular objects can be divided into closed polygons and smooth closed curves. According to the type of the object, the piecewise linear interpolation method or the cubic spline interpolation method is used to convert the edge contour coordinates of the shape of the irregular object into a parametric equation , which is used to describe the contour information of the irregular object, that is, the actual action point at the end of the mobile robot link; let the parameter t range be, where t = 0 corresponds to the starting point, t = 1 corresponds to the end point, that is, returning to the starting point.

[0057] Specifically, S201. If the object type is a closed polygon, then by dividing the interval into nDivide it equally. Each sub-interval corresponds to the line segment between two adjacent points. Calculate the edge coordinate values within each sub-interval using linear interpolation. The specific form of the parametric equation is as follows:

[0058] For the i th line segment , when , there is

[0059] ,

[0060] ,

[0061] where and represent the i th corner point of the closed polygon x coordinate and y coordinate. When , to ensure closure.

[0062] S202. If the object is not a closed polygon but a smooth closed curve, use cubic spline interpolation to represent it. By constructing piecewise cubic polynomials, ensure that the first derivative and the second derivative are continuous at the connection points of adjacent segments, thus forming an overall smooth curve. In the parameter allocation, first allocate according to the cumulative distance between adjacent points, , for each interval the corresponding cubic polynomial is as follows:

[0063] ,

[0064] ,

[0065] where the coefficients are solved and determined by satisfying the following conditions:

[0066] (a) Position continuity ;

[0067] (b) First derivative continuity ;

[0068] (c) Second derivative continuity ;

[0069] (d) Periodic boundary conditions .

[0070] S300: Expand the shape of the object to generate a set of reachable position points for the robot, and establish the mapping relationship between the parametric equation and the reachable positions.

[0071] Specifically, expand the object shape described by the parametric equation along the normal direction, and the expansion length is the length of the robot linkL , the unit normal vector of the object shape is , establish the mapping relationship between the parametric equation and the reachable position as , and obtain the candidate two-dimensional coordinate points that the robot can directly reach.

[0072] In the actual dilation process, due to the non-convexity of the object shape, it may cause the dilation points to be inside the object or unreachable after considering the robot volume, and these points need to be removed; the non-smooth points in the object shape parametric equation cannot be directly applied in practice, and there are no dilation points at these points; the particularity of the object shape may result in the situation where two or more edge points correspond to the same dilation point. For simplicity, it is considered that after the mobile robot reaches this dilation point, fixing these two or more edge points has the same effect, and finally, the set of actual reachable position points after removal is obtained S .

[0073] S400: Construct a multi-objective optimization function, use the coordinate points as the variables to be optimized, optimize the quality of the initial population using the improved NSGA-II algorithm and define the constraint violation degree, and then obtain the optimal solution. According to the number of robots N determine the coordinates of the best action points. The specific steps are as follows:

[0074] S401. First, construct a multi-objective function, specifically:

[0075] (1) According to the corresponding position of each robot as , the position corresponding to the centroid of the object is , construct a multi-objective optimization function as:

[0076]

[0077] where, .

[0078] (2) Position dispersion objective: Maximize the distance between robots to improve the stability of the multi-mobile robot collaborative handling system, and establish a position dispersion objective function as .

[0079] (3) Torque maximization objective: Make each mobile robot as far away as possible from the centroid of the irregular object (assuming the object has a uniform mass distribution) to improve the upper limit of torque output, and establish a torque maximization objective function as .

[0080] (4) Shape closure objective: Ensure that the acting forces are symmetrically distributed. Assume that the upper limit of the force provided by the robots is the same, and when receiving forces in the same direction, they can achieve independent translation without rotation, and establish a shape closure objective function as .

[0081] (5) Incorporate the collision constraints between mobile robots. The distance between robots needs to meet the safety distance threshold. where R is the link length. is the safe elastic distance.

[0082] S402. Optimize the quality of the initial population using the improved NSGA-II algorithm and define the constraint violation degree, and then obtain the optimal solution.

[0083] Based on the constructed multi-objective function and collision constraints, solve the corresponding Pareto front in the optimization problem in the improved NSGA-II algorithm.

[0084] In this embodiment, use the path parameter in the parametric equation as the decision variable, and generate the initial population through real number coding. Initialize the population size, crossover probability, and mutation probability in the population, solve the optimal solution of the single-objective function, and use this as the high-quality initial individuals in the initial population to reduce the number of iterations and accelerate the solution efficiency.

[0085] For the constraint problem, define the constraint violation degree. . Define the dominance relationship in non-dominated sorting. In the hierarchical comparison, the feasible solutions are always better than the infeasible solutions . The infeasible solutions are sorted in ascending order by value. If two solutions and are both feasible solutions, and is not inferior to on all objectives and is strictly better on at least one objective, then dominates ; if both solutions are infeasible, the smaller solution is better. According to the above sorting rules, layer by layer according to the Pareto dominance relationship, classify the non-dominated solutions into the first layer Front1, and the remaining after removing the non-dominated solutions are classified into the second layer Front2, and so on. When calculating the crowding degree, only consider the distribution of feasible solutions to ensure the diversity of feasible solutions. When merging the parent and offspring, preferentially eliminate solutions.

[0086] Iteratively obtain the feasible Pareto front. All solutions satisfy the constraint conditions of the objective function. Finally, assign weights to the position dispersion, torque maximization, and form closure objective functions according to the actual task requirements, and determine the optimal solution through weighted summation , that is N the best reachable position coordinates and the best action point coordinate information of

[0087] In the MATLAB platform, the above-mentioned multi-objective functions and constraints are designed for solution. In the population initialization optimization, high-quality initial individuals are generated by combining single-objective optimization pre-solutions. By setting the population size to 50, the crossover probability to 0.7, and the mutation probability to 0.4, first, the optimal solutions of the single-objective function are obtained through 3 * 200 iterations, which are used as high-quality individuals in the initial population of multi-objective optimization. Then, the Pareto front solutions are screened through 200 iterations. Based on the actual task requirements, the weights of indicators such as position dispersion, torque upper limit, and form closure are assigned, and the final deployment positions of N robots are determined through weighted summation. As Figure 2 shown in the edge extraction and action point optimization results of different-shaped objects, three typical target shapes ( Figure 2 the rectangular-like object in (a), Figure 2 the L-shaped object in (b), and Figure 2 the elliptical-like object in (c) can be regarded as regular objects, regular polygon objects, and irregular objects respectively) are tested, and the improved algorithm can efficiently converge to the optimized position configuration that satisfies geometric constraints.

[0088] S500: After obtaining the optimal action point coordinates, control multiple robots to perform rigid formation to complete the collaborative handling task of irregular objects.

[0089] After obtaining the optimal action point coordinates, design a Mecanum wheel mobile robot equipped with a passive single-degree-of-freedom rotating link. Rigidly connect the end of the link to the object. Through the DMPC controller with rigid formation hard constraints, realize trajectory tracking, adjust the formation attitude in real time, and control the mobile robot to complete the collaborative handling task of irregular objects to ensure the dynamic stability of the system during the handling process. The results show that the system exhibits high robustness and effectiveness in handling objects with complex shapes.

[0090] The method proposed in this implementation example obtains the original image through the installed RGB camera, uses image processing technology to obtain the set information of the contour coordinate points of the irregular object, and obtains the mapping relationship between the parameters and the contour coordinate points by establishing a parametric equation; according to the link lengths in the designed mobile robot handling mechanism, expand the object contour outward along the normal direction to obtain the set of coordinate points where the mobile robot can exert an action and reach, and at the same time obtain the mapping relationship between the parameters and the reachable coordinate points; design a multi-objective optimization function that takes into account handling safety and stability, the upper limit of handling, and efficiency. Using the improved NSGA-II algorithm, take the object parametric equation and the number of robots as inputs. First, iterate to obtain the optimal solutions of each objective function as individuals in the initial population, then iterate to solve for the Pareto front, and according to the actual handling needs, adjust the weights of the three objective functions on the Pareto front to obtain the optimal solution, and then the optimal reachable position coordinates and optimal action point coordinate information of multiple robots can be obtained.

[0091] Embodiment 2

[0092] In a typical embodiment of the present invention, this embodiment discloses a multi-robot collaborative handling system for irregular objects based on multi-objective optimization, including:

[0093] An edge contour coordinate extraction module, configured to: obtain the shape information of the object and extract the edge contour coordinates of the object;

[0094] A parametric equation construction module, configured to: convert the edge contour coordinates of the object into parametric equations of the actual action points at the ends of the robot linkages;

[0095] An accessible position point set generation module, configured to: expand the shape of the object, generate a set of accessible position points for the robot, and establish a mapping relationship between the parametric equations and the accessible positions;

[0096] A multi-objective optimization module, configured to: construct a multi-objective optimization function, use the coordinate points as variables to be optimized, optimize the quality of the initial population using an improved NSGA-II algorithm and define the constraint violation degree, and then obtain the optimal solution, and determine the coordinates of the best action points according to the number of robots;

[0097] A task execution module, configured to: after obtaining the coordinates of the best action points, control the multi-robots to perform rigid formation to complete the collaborative handling task of the irregular object.

[0098] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A multi-robot collaborative handling method for irregular objects based on multi-objective optimization, characterized in that: include: Obtain the shape information of the object and extract the edge contour coordinates of the object; Convert the edge contour coordinates of the object into the parametric equation of the actual action point at the end of the robot link; Expand the shape of the object, establish the mapping relationship between the parameter equation and the reachable position, and obtain the actual reachable position point set of the robot; Construct a multi-objective optimization function, take the coordinate points as the variables to be optimized, use the improved NSGA-II algorithm to optimize the initial population quality and define the constraint violation degree, and then find the optimal solution, and determine the coordinates of the best action point according to the number of robots; After obtaining the coordinates of the optimal action point, multiple robots are controlled to form a rigid formation to complete the collaborative handling task of irregular objects.

2. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 1, characterized in that: The method of obtaining the shape information of the object and extracting the edge contour coordinates of the object is specifically as follows: obtaining the original image of the object through an RGB camera, converting the original image into a grayscale image and performing Gaussian filtering, and then extracting the edge contour coordinates of the object through an edge detection algorithm.

3. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 1, characterized in that: The method of converting the edge contour coordinates of the object into the parameter equation of the actual action point of the robot connecting rod end is specifically: according to the object type, using the corresponding interpolation method, the edge coordinates of the object are converted into the parameter equation of the actual action point of the robot connecting rod end.

4. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 3, characterized in that: The object types include closed polygons and smooth closed curves.

5. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 4, characterized in that: If the object type is a closed polygon, the parametric equation is constructed using piecewise linear interpolation; If the object type is a smooth closed curve, the cubic spline interpolation method is used to construct the parametric equation.

6. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 1, characterized in that: Expand the object shape, establish the mapping relationship between the parametric equation and the reachable position, and obtain the set of actual reachable position points of the robot. Specifically, expand the object shape described by the parametric equation outward along the normal direction. The expansion length is the length of the robot connecting rod. Establish the mapping relationship between the parametric equation and the reachable position, and obtain the candidate two-dimensional coordinate points that the robot can directly reach. Eliminate invalid points caused by non-convexity or volume conflict of the object, and finally obtain the set of actual reachable position points after elimination.

7. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 1, characterized in that: The construction of the multi-objective optimization function is specifically as follows: A multi-objective optimization function is constructed based on the corresponding position of each robot and the position corresponding to the center of mass of the object; the position dispersion objective function is established by maximizing the distance between robots; the torque maximization objective function is established based on keeping each robot as far away from the center of mass of irregular objects as possible; the shape closure objective function is established based on the symmetrical distribution of the force; collision constraints are added between robots so that the distance between robots meets the safety distance threshold.

8. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 7, characterized in that: Based on the constructed multi-objective functions and collision constraints, the improved NSGA-II algorithm is used to take the object parameter equation and the number of robots as input. The optimal solution of each objective function is first iterated as the individual in the initial population, and then the Pareto front is obtained by iterative solution. According to the actual task requirements, the weights of the position dispersion objective function, the moment maximization objective function and the shape closure objective function are adjusted on the Pareto frontier. The optimal solution is determined by weighted summation, and the optimal reachable position coordinates and the optimal action point coordinates of the multiple robots are obtained.

9. The method for multi-robot collaborative handling of irregular objects based on multi-objective optimization as claimed in claim 1, characterized in that: After obtaining the coordinates of the optimal point of action, the end of the robot's connecting rod is rigidly connected to the object, and the formation posture is adjusted in real time through a controller with rigid formation hard constraints to control multiple robots to complete the collaborative handling task of irregular objects.

10. Based on multi-objective optimization, a multi-robot collaborative handling system for irregular objects is characterized by: include: The edge contour coordinate extraction module is configured to: obtain shape information of the object and extract edge contour coordinates of the object; The parametric equation building module is configured to: transform the edge contour coordinates of the object into the parametric equation of the actual action point of the robot link end; The reachable position point set generation module is configured to: expand the object shape, establish a mapping relationship between the parameter equation and the reachable position, and obtain the actual reachable position point set of the robot; The multi-objective optimization module is configured to: construct a multi-objective optimization function, take the coordinate points as the variables to be optimized, use the improved NSGA-II algorithm to optimize the initial population quality and define the constraint violation degree, and then obtain the optimal solution, and determine the coordinates of the best action point according to the number of robots; The task execution module is configured to: after obtaining the coordinates of the optimal action point, control multiple robots to form a rigid formation to complete the collaborative handling task of irregular objects.

Images