Multi-target grabbing-oriented mobile operation robot continuous motion planning method

Abstract

The invention provides a mobile operation robot continuous motion planning method for multi-target grabbing, and belongs to the technical field of target detection, grabbing and man-machine cooperation interaction of mobile operation robots. According to the method, a kinematic model of the mobile operation robot is established, then an effective parking area is determined, and the area is a circle with each grabbing target point as the circle center and the operation range of a mechanical arm as the radius; and selecting a plurality of candidate stop points on the circle, calculating the shortest path of the robot among the candidate stop points of each target area, and introducing a steering angle constraint to determine an optimal stop point and an access sequence so as to realize efficient continuous motion planning in the multi-target scene. And when the robot reaches the effective parking area of each target, identifying a target object through target detection based on a depth camera. After target detection is completed, the robot can adjust the posture and the position of the mechanical arm according to the detected target position, so that the mechanical arm can smoothly complete grabbing operation in an effective working area.

Description

Technical Field

[0001] The present invention belongs to the technical field of target detection, grasping and human-machine collaborative interaction of mobile working robots, and relates to a continuous motion planning method of a mobile working robot for multi-target grasping. Background Art

[0002] The mobile operation robot is a composite robot system. The motion system consists of a mobile chassis and a six-degree-of-freedom robotic arm, and the perception system is mainly composed of a three-dimensional laser radar and a depth camera. The mobile chassis of the mobile operation robot can adopt a wheel-foot type, a crawler type, or even a composite drive mode combining the two to ensure flexible movement in a wide range of working environments. The robotic arm installed on it can work in conjunction with the mobile chassis to complete the grasping of the target object, forming an overall operation system. By combining intelligent sensors with the six-degree-of-freedom robotic arm, the mobile operation robot can realize environmental perception, autonomous decision-making and motion control, and has the ability to move and operate autonomously. Compared with traditional fixed-position operations of robotic arms and ordinary mobile robots, this robot can realize the joint motion planning of the chassis and the robotic arm, so as to efficiently grasp the target object in a larger workspace. This flexibility makes the mobile operation robot have a wide range of application potential in many fields such as logistics handling, industrial production and orchard picking.

[0003] The literature (Fan Jiangbo. Research on the operation control of a harvesting robot based on an embedded system [J]. Agricultural Mechanization Research.) proposed an operation control method for a strawberry harvesting robot, aiming to solve the problems of backward means, low precision and low efficiency in traditional harvesting operations. Based on an in-depth analysis of the structure and working principle of the harvesting robot, this study introduced an embedded system and completed the kinematic analysis of the harvesting manipulator. However, this method mainly focuses on the picking of a single target and is not suitable for continuous motion planning in multi-target grasping scenarios. It lacks systematic consideration of the coordinated motion of the mobile robot and the manipulator. Therefore, it is impossible to effectively handle complex tasks or achieve efficient object manipulation. The literature (Liu Zhenyu. Research on motion control of a quadruped dual-arm mobile operation robot [D]. Supervisor: Zhang Qin. Jinan University, 2023.) discusses the application of a quadruped mobile robot equipped with dual manipulators in industrial production, focusing on the motion continuity and coordination of the robot in complex environments. This study uses a sampling-based planning method to realize the path planning of the clamped object and the torso, and performs kinematic analysis to obtain the path in the joint space. For kinematic analysis, the kinematic models of the legs, robotic arms and trunk were established, and an inverse kinematics solution method based on the Mayfly Algorithm (MA) was designed. The kinematic equations were established and solved by improving the DH method. However, this method is too complex. In the face of multi-target continuous motion planning scenarios, the solution execution efficiency is low and the process is cumbersome. It is difficult to be well applied in actual working environments. Based on this, the present invention proposes a continuous motion planning method for a mobile working robot for multi-target grasping. Summary of the invention

[0004] In order to overcome the defects and shortcomings of existing mobile working robots in the face of multi-target continuous motion planning and grasping, the present invention proposes a method for continuous motion planning of mobile working robots for multi-target grasping. The method first establishes a kinematic model for the mobile working robot to prepare for the subsequent research on the continuous motion planning of the mobile working robot, wherein the mobile chassis adopts the Ackerman motion model, and the robotic arm uses the improved DH method to establish the kinematic model; then facing the continuous motion planning of the mobile working robot in a multi-target scenario, the effective docking area of ​​the mobile working robot is first determined, which is a circle with the grasping target point as the center and the operable range of the robotic arm as the radius, and a stop is selected on the circle in the effective docking area. Several points are candidate docking points. Facing several candidate docking points on each target area, a genetic algorithm is used to calculate the shortest path of the mobile robot with multiple target points, and the optimal docking point is obtained to determine the optimal access sequence of the mobile robot facing the continuous motion planning of the multi-target scene; when the mobile robot reaches the effective docking area of ​​each target, target detection based on a depth camera is used. After the target detection is completed, the posture and position of the robot arm are adjusted according to the detected target coordinates so that it can achieve the grasping operation in the effective working area; finally, in the continuous motion planning process of the multi-target grasping operation, in order to cope with potential obstacles or environmental changes, the point cloud information of the laser radar can be obtained in real time. When the laser radar detects a new obstacle, it avoids the obstacle to ensure that the robot can reach the next target area safely and smoothly. This real-time path optimization capability enables the robot to quickly adapt when encountering obstacles or other unforeseen situations, improving the overall response speed of the system and the efficiency of task completion.

[0005] The technical solution of the present invention:

[0006] A continuous motion planning method for a mobile working robot for multi-target grasping includes the following steps:

[0007] S1. Establish a kinematic model for the mobile working robot;

[0008] S2, continuous motion planning of mobile working robots, including determining the effective docking area of ​​the mobile working robot, which is a circle with the grasping target point as the center and the manipulatable range of the robot arm as the radius; selecting candidate docking points on the circle of each effective docking area; calculating the shortest path of the mobile working robot with multiple target points, and introducing steering angle constraints to obtain the best docking point, and determining the optimal access sequence of the mobile working robot facing the continuous motion planning of the multi-target scene;

[0009] S3. The mobile working robot performs multi-target grasping tasks based on the optimal access sequence and the best stop point.

[0010] Furthermore, in step S1, the mobile chassis of the mobile working robot adopts the Ackerman motion model, which is characterized in that all wheels point to the same instantaneous turning center when turning, so that the turning radius of the front and rear wheels is consistent, reducing the lateral slip of the tires; assuming the turning angle δ of the robot, then:

[0011]

[0012] Among them, R represents the turning radius, W represents the distance between the left and right wheels, and L represents the distance between the front and rear wheel axles;

[0013] The kinematic modeling of a six-degree-of-freedom robotic arm is the basis for realizing the control and path planning of the end effector of the robotic arm. By using the modified-DH parameter method, the relationship between the joints and connecting rods of the robotic arm can be expressed in a simple mathematical form, and then the kinematic model of the robotic arm can be established.

[0014] Furthermore, in a multi-target grasping task, the effective docking area refers to the area within each target area where the robot can safely and effectively reach and complete the grasping operation. The definition of this area needs to comprehensively consider factors such as the robot's operating range, accuracy requirements, and obstacle locations. The method for determining the effective docking area is as follows:

[0015] Define T i ={t1,t2,…,t n}, where n is the specific number of multiple targets, and the effective working radius of the robot is defined as r; then the effective docking area should be within the effective working radius r of the robot. Specifically, for each target point t i All points within the radius r are considered as potential docking areas. The mobile robot uses constraints to ensure that the entire grabbing target area can be covered at any docking position and will not fail to perform tasks due to exceeding the operating range. Based on the above description, the following definition can be obtained:

[0016] Z i ={(x,y)|(x i ) 2 +(yy i ) 2 ≤r 2}

[0017] Among them, Z i Represents the target t i The effective docking area (x i ,y i ) is the target point t i The coordinates of the robot arm are shown in Figure 2, and r is the radius of the effective docking area, that is, the effective working radius of the robot arm.

[0018] Furthermore, the candidate docking point refers to a predetermined position where the mobile working robot can choose to dock and perform a grasping operation during the execution of the task. The motion planning of the mobile working robot is not only about connecting the target points, but also about how to efficiently reach the specific location in the target area. The candidate docking point provides multiple optional docking positions for the path planning algorithm to optimize and select the shortest or optimal path. By generating the candidate docking points, it can be ensured that the selected point is within the effective working range of the robot arm and can successfully perform the grasping task.

[0019] The generation method of candidate stopping points depends on the characteristics of the target area, the working radius of the robot system, and the requirements for path planning and accuracy. In this task, the robot's target area is a circle, and the candidate points can be evenly distributed on the boundary of the target area.

[0020] Set target t i The effective docking area Z i The center position t i is (x i ,y i ), generate m candidate stop points, evenly distributed on the circumference, for each candidate stop point c ij (j=1,2,…,m):

[0021]

[0022] c ij =(x i +r·cos(θ j ),y i +r·sin(θ j ))

[0023] Among them, θ j represents the angle when generating the jth candidate stop point. When j starts from 1, the angle is evenly distributed from 0 to 2π through (j-1); c ij Represents the target t i The effective docking area Z i The position of the jth candidate docking point; r represents the radius of the effective docking area of ​​the target (i.e., the working radius of the robot arm); cos(θ j ) and sin(θ j ) is used to calculate the horizontal and vertical coordinate offsets of the candidate stop points on the circumference.

[0024] Through the above formula, m stop points can be evenly generated on the boundary of the effective stop area of ​​each target. These points will be used as candidate points for path planning to select the best access order and determine the optimal path.

[0025] Furthermore, the optimal access sequence of the mobile robot for continuous motion planning in multi-target scenarios is determined as follows:

[0026] Problem definition: Given n targets with valid docking areas Z1, Z2, …, Z n , each valid docking area Z i There are m candidate stops c i1 ,c i2 ,…,c im ; The goal is to find the optimal visit sequence and the best stop point for each valid stop area, so that the total path length is the shortest and the smoothness and stability of the robot path are ensured; For this problem, a genetic algorithm can be used to solve it, the steps are as follows:

[0027] 1) Initialization

[0028] Each individual is encoded by a combination of a visit sequence and a corresponding candidate stop point; the encoding form can be an array, where each position represents the number of a target valid stop area and specifies the stop point index in the area; for example, the sequence {c 11 ,c 22 ,…,c nm}, where c nm Represents the target t n The effective docking area Z n The mth candidate stop point is selected; then a set of initial sequences is randomly generated, each sequence individual contains a random visit order of the target area and the corresponding stop point selection.

[0029] 2) Define the objective function

[0030] The objective function is a fitness function, which measures the quality of the path. The shorter the path, the higher the fitness. In the genetic algorithm, the shorter the total path length, the better. Therefore, the total path length can be used as one of the constraints of the objective function. Therefore, we have:

[0031]

[0032] Where n represents the total number of valid docking areas; d(c iu ,c (i+1)v ) represents the selected stop point c of the i-th valid stop area iu and the selected stop point c of the i+1th valid stop area (i+1)v The Euclidean distance between them; the last item d(c nu ,c 1v ) indicates the last stop point c in the valid stop area nu Stop c with the first valid stop area 1v The distance between them; d(c iu ,c(i+1)v ) can be calculated by the following formula:

[0033]

[0034] In order to ensure the continuity of the path and avoid sharp turns or non-smooth curves to reduce the non-smoothness of the trajectory, it is also necessary to introduce a steering angle constraint; for each path d(c iu ,c (i+1)v ), the steering angle Δδ needs to be calculated:

[0035]

[0036] Δδ i =|δ curr -δ prev |

[0037] Among them, δ curr Indicates the direction angle of the current segment, δ prev Indicates the direction angle of the previous segment, Δδ i Indicates the turning angle of this path;

[0038] Therefore, the objective function needs to integrate the smoothness of path length and steering angle, defined as:

[0039]

[0040] Among them, α and β are weight parameters used to adjust the relative importance of path length and turning smoothness; is the sum of all steering angles;

[0041] The smaller the value of Total Cost, the higher the fitness. Therefore, the fitness can be defined as the inverse of TotalCost:

[0042]

[0043] 3) Select the best individual

[0044] A probability-based selection method is used, and individuals with higher fitness are more likely to be selected; accordingly, for each individual h, its selection probability p is calculated h :

[0045]

[0046] Among them, n is the size of the sequence, Fitness h is the fitness value of the hth individual; during selection, an elite retention strategy can be used to retain the individuals with the highest fitness directly to the next generation to prevent the optimal solution from being lost.

[0047] 4) Use crossover mutation operation to avoid falling into local optimal solution

[0048] The crossover operation generates new individuals by combining the excellent features of the parent individuals, so that the new solution may have a shorter path; crossover helps to explore new solution spaces and accelerate convergence to the global optimal solution. The purpose of the mutation operation is to introduce new information, prevent the algorithm from falling into the local optimal solution, and maintain the diversity of the solution space.

[0049] 5) If one of the following conditions is met, the algorithm stops:

[0050]

[0051] Among them, q represents the current number of iterations, N max Indicates the maximum number of iterations; |Fitness(q)-Fitness(qk)|<∈ means that if the fitness of the current iteration changes less than the fitness of the kth generation ago by less than the threshold ∈, then stop; Fitness best ≥Fitness target Indicates that if the fitness value of the optimal individual reaches the target fitness target , then stop; finally, if the algorithm running time exceeds the maximum running time T max , then stop.

[0052] Furthermore, in step S3, the mobile operation robot performs multi-target grasping work, which specifically includes:

[0053] Use Python's Matplotlib to generate a working roadmap for multi-target continuous motion planning of a mobile robot. This working roadmap covers the optimal sequence and optimal docking points for the robot to visit multiple target effective docking areas, as well as the minimization design of the path.

[0054] Based on the work roadmap, the mobile working robot goes to the effective docking area of ​​each target in turn and performs the grasping task; each time it reaches the effective docking area of ​​a target, the robot locates the target object according to the optimal docking point position and the visual data obtained by the depth camera, and adjusts the grasping position and posture in real time to ensure accurate grasping operation.

[0055] Furthermore, in order to cope with potential obstacles or environmental changes, the mobile working robot obtains point cloud information of the laser radar in real time when performing multi-target grasping work; when the laser radar detects an obstacle, it avoids the obstacle to ensure that the robot can reach the next target area safely and smoothly.

[0056] Beneficial effects of the present invention: The present invention proposes a continuous motion planning method for a mobile working robot for multi-target grasping tasks. The method is based on the kinematic model of the mobile robot and the manipulator. First, the effective docking area of ​​the target area is defined, and multiple candidate docking points are selected in the area. Combined with the path minimization goal, the steering angle constraint is introduced to optimize the access order of the robot between multiple docking points, and finally the optimal path is obtained. This path planning process not only takes into account the distance between the targets, but also ensures the smoothness and stability of the robot path by controlling the change of the steering angle, avoiding sharp turns and discontinuous trajectories. After experimental verification, the method performs well in optimizing the robot's multi-target access order and path length. Thanks to the high precision and high efficiency of the method, the mobile robot can adjust the motion trajectory and grasping operation in real time when performing multi-target grasping tasks. Combined with the high-precision target detection function of the laser radar and the depth camera, better pose estimation and efficient multi-target grasping capabilities are achieved, which is particularly suitable for task scenarios such as industrial manufacturing, logistics distribution and unmanned inspection in dynamic and complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a workflow diagram.

[0058] Figure 2 This is a simulation diagram of a mobile working robot.

[0059] Figure 3 It is the coordinate relationship diagram between two adjacent connecting rods.

[0060] Figure 4 This is a diagram of the effective docking area for the mobile working robot.

[0061] Figure 5 Continuous motion planning roadmap for mobile robots for multi-object grasping.

[0062] Figure 6 It is the optimal grasping path for the mobile working robot.

[0063] Figure 7 It is the curve of fitness change during the solution process. DETAILED DESCRIPTION

[0064] The specific implementation of the present invention is described in detail below in conjunction with the technical scheme and the accompanying drawings.

[0065] The mobile chassis used in the specific implementation process of the present invention is an Ackerman motion model. Through appropriate steering angle design, it is ensured that all wheels point to the same instantaneous steering center when turning, so that the turning radius of the front and rear wheels is consistent, and the lateral slip of the tires is reduced. The maximum linear speed of the mobile chassis is 2m / s, and the maximum turning angle of the front wheel is + / -30 degrees; the 6-axis robot arm has 6 degrees of freedom, a body weight of 10kg, a rated load of 3kg, a working radius of 900mm, a repeatability accuracy of ±0.5mm, and joint operation speeds: J1-J2 is 180° / s, J3-J6 is 225° / s, where J s Represents the sth joint of the robotic arm.

[0066] The first step is to determine the model parameters of the mobile chassis and robotic arm.

[0067] According to the steering angle δ of the left and right front wheels of the mobile chassis l and δ r , we can determine the distance R from the instantaneous steering center to the rear axle, the distance between the front and rear axles L = 52cm, and the width between the left and right wheels (wheel spacing) W = 40cm. For a 6-DOF manipulator, based on the parameter information of the manipulator, the MD-H parameters of the manipulator can be obtained as follows:

[0068]

[0069] Based on the above relationship, the kinematic model of the robot arm can be determined.

[0070] The second step is to solve the continuous motion planning of the mobile working robot for multi-target grasping.

[0071] First, define the effective docking area. Here we take five targets as an example. There are T i ={t1,t2,t3,t4,t5}, ​​where t1 = 0,0], t2 = [5,0], t3 = [5,5], t4 = [0,5], t5 = [0,8]. The effective working radius of the robot arm is r = 1m. Based on the above description, the following definition can be obtained:

[0072] Z i ={(x,y)|(x i ) 2 +(yy i ) 2 ≤1 2}

[0073] Where Z i Represents the target t i The effective docking area (x i ,y i ) is the target point t i The valid docking area is visible Figure 4 .

[0074] Then generate candidate stops. i The effective docking area Z i The center position t i is (x i ,y i ), generate m = 4 candidate stop points, distributed on the circumference, for each candidate point c ij (h=1,2,3,4):

[0075]

[0076] c ij =(x i +r·cos(θ j ),y i +r·sin(θ j ))

[0077] Based on the above two formulas, we can get the candidate stop points corresponding to each target point, such as Figure 4 shown.

[0078] Then, the optimal access sequence of multiple targets is determined based on the genetic algorithm. During encoding and initialization, each individual is encoded by a combination of an access sequence and a corresponding stop point, such as the sequence {c 11 ,c 22 ,…,c nm}, where c nm Represents the target t n The effective docking area Z n The mth candidate stop. The fitness function is defined as:

[0079]

[0080] Among them, Total Cost:

[0081]

[0082] Among them n=5, m=4, α=0.7, β=0.3; d(c iu ,c (i+1)v ) represents the selected stop point c of the i-th valid stop area iu and the selected stop point c of the i+1th valid stop area (i+1)v The Euclidean distance between nu ,c 1v ) indicates the last stop point c in the valid stop area nu Stop c with the first valid stop area 1v The distance between them; Euclidean distance d(ciu ,c (i+1)v ) can be calculated by the following formula:

[0083]

[0084] Then the roulette wheel selection strategy is used to determine the selection probability of each individual, with a mutation probability of 0.1, a crossover probability of 0.9, a number of iterations of 200, and a maximum running time of T. max Take 5s, target fitness target Take 0.6, but stop iterating if one of the following conditions is met:

[0085]

[0086] Through this method, a roadmap for continuous motion planning of mobile robot for multi-target grasping can be obtained, such as Figure 5 As shown, the optimal path can also be obtained, such as Figure 6 ;Fitness change curve diagram see Figure 7 .

[0087] The third step is to perform the grasping work of the robotic arm based on the work roadmap of the mobile working robot.

[0088] The work route map of the mobile robot's multi-target continuous motion planning covers the optimal order and optimal stop points for the robot to visit multiple target areas. Based on the generated work route map, the mobile robot goes to each target area in turn and performs the grasping task. When arriving at each target area, the robot will locate the target object based on the optimal stop point position and the visual data obtained by the depth camera, and adjust the grasping position and posture in real time to obtain the position T of the target object. obj-in-cam After that, it is necessary to convert it to the robot arm coordinate system through coordinate transformation, using the eye-on-hand method to define T boj-in-base is the position of the object in the base coordinate system, T hand-in-base is the hand position in the base coordinate system, T cam-in-hand is the position of the camera in the terminal coordinate system, T obj-in-cam is the position of the object in the camera. After calibration, we can get:

[0089]

[0090] as well as:

[0091]

[0092] Substitute the above matrix into the following formula:

[0093] T obj-in-base =T hand-in-hase ·Tcam-in-hand ·T obj-in-cam

[0094] At this point, the mobile working robot can perform precise grasping operations when moving to the optimal target stop point and complete the grasping task in multi-target continuous motion planning.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A continuous motion planning method for a mobile robot for multi-target grasping, characterized in that: The following steps are involved: S1. Establish a kinematic model for the mobile working robot; S2, continuous motion planning of mobile working robots, including determining the effective docking area of ​​the mobile working robot, which is a circle with the grasping target point as the center and the manipulatable range of the robot arm as the radius; selecting candidate docking points on the circle of each effective docking area; calculating the shortest path of the mobile working robot with multiple target points, and introducing steering angle constraints to obtain the best docking point, and determining the optimal access sequence of the mobile working robot facing the continuous motion planning of the multi-target scene; S3. The mobile working robot performs multi-target grasping tasks based on the optimal access sequence and the best stop point.

2. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 1, characterized in that: In step S1, the mobile chassis of the mobile working robot adopts the Ackerman motion model, which is characterized in that all wheels point to the same instantaneous turning center when turning, so that the turning radius of the front and rear wheels is consistent, reducing the lateral slip of the tires; assuming the robot's turning angle δ, then: Among them, R represents the turning radius, W represents the distance between the left and right wheels, and L represents the distance between the front and rear wheel axles; The kinematic model of the six-degree-of-freedom robotic arm is established using the modified-DH parameter method.

3. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 2, characterized in that: The method for determining the effective docking area is as follows: Define T i ={t1,t2,…,t n }, where n is the specific number of multiple targets, and the effective working radius of the robot arm is defined as r; i All points within a radius r are considered as potential docking areas. The valid docking areas are as follows: Z i ={(x,y)|(xx i ) 2 +(yy i ) 2 ≤r 2 } Among them, Z i Represents the target t i The effective docking area (x i ,y i ) is the target point t i The coordinates of the robot arm are shown in Figure 2, and r is the radius of the effective docking area, that is, the effective working radius of the robot arm.

4. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 3 is characterized in that: The method for selecting the candidate stop point is: Set target t i The effective docking area Z i The center position t i is (x i ,y i ), generate m candidate stop points, evenly distributed on the circumference, for each candidate stop point c ij , j=1,2,…,m: c ij =(x i +r·cos(θ j ),y i +r·sin(θ j )) Among them, θ j represents the angle when generating the jth candidate stop point. When j starts from 1, the angle is evenly distributed from 0 to 2π through (j-1); c ij Represents the target t i The effective docking area Z i The position of the jth candidate docking point; r represents the radius of the effective docking area of ​​the target, i.e., the working radius of the robot; cos(θ j ) and sin(θ j ) is used to calculate the horizontal and vertical coordinate offsets of the candidate stop points on the circumference.

5. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 4, characterized in that: When determining the optimal visit sequence of a mobile robot facing a multi-target continuous motion planning scenario, the problem is defined as: given n valid parking areas Z1, Z2, …, Z n , each valid docking area Z i There are m candidate stops c i1 ,c i2 ,…,c im ; The goal is to find the optimal visit sequence and the best stop point for each valid stop area, so that the total path length is the shortest and the smoothness and stability of the robot path are ensured.

6. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 5, characterized in that: Genetic algorithm is used to solve the optimal visit sequence and the best stop point for each valid stop area, including: 1) Initialization Each individual is encoded by a combination of a visit sequence and a corresponding candidate stop point; the encoding form can be an array, in which each position represents the number of a target valid stop area and specifies the stop point index of the area; then a set of initial sequences is randomly generated, and each sequence individual contains a random visit sequence of the target area and the corresponding stop point selection; 2) Define the objective function The objective function is a fitness function, which measures the quality of the path. The shorter the path, the higher the fitness. The total length of the path is used as one of the constraints of the objective function, so: Where n represents the total number of valid docking areas; d(c iu ,c (i+1)v ) represents the selected stop point c of the i-th valid stop area iu and the selected stop point c of the i+1th valid stop area (i+1)v The Euclidean distance between nu ,c 1v ) indicates the last stop point c in the valid stop area nu Stop c with the first valid stop area 1v The distance between In order to ensure the continuity of the path and avoid sharp turns or non-smooth curves to reduce the non-smoothness of the trajectory, it is also necessary to introduce a steering angle constraint; for each path d(c iu ,c (i+1)v ), the steering angle Δδ needs to be calculated: Dd i =|δ curr -d prev | Among them, δ curr Indicates the direction angle of the current segment, δ prev Indicates the direction angle of the previous segment, Δδ i Indicates the turning angle of this path; Therefore, the objective function needs to integrate the smoothness of path length and steering angle, defined as: Among them, α and β are weight parameters used to adjust the relative importance of path length and turning smoothness; is the sum of all steering angles; The smaller the value of Total Cost, the higher the fitness; therefore, the fitness can be defined as the inverse of Total Cost: 3) Select the best individual A probability-based selection method is used, and individuals with higher fitness are more likely to be selected; accordingly, for each individual h, its selection probability p is calculated h : Among them, n is the size of the sequence, Fitness h is the fitness value of the hth individual; 4) Use crossover mutation operation to avoid falling into local optimal solution; 5) If one of the following conditions is met, the algorithm stops: Among them, q represents the current number of iterations, N max Indicates the maximum number of iterations; |Fitness(q)-Fitnerr(qk)|<∈ means that if the fitness of the current iteration changes less than the fitness of the kth generation ago by less than the threshold ∈, then stop; Fitness best ≥Fitness target Indicates that if the fitness value of the optimal individual reaches the target fitness target , then stop; finally, if the algorithm running time exceeds the maximum running time T max , then stop.

7. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 1, characterized in that: In step S3, the mobile operation robot performs multi-target grasping work, which specifically includes: Use Python's Matplotlib to generate a working roadmap for multi-target continuous motion planning of a mobile robot. This working roadmap covers the optimal sequence and optimal docking points for the robot to visit multiple target effective docking areas, as well as the minimization design of the path. Based on the work route map, the mobile working robot goes to the effective parking area of ​​each target in turn and performs the grasping task; Every time the robot reaches the effective docking area of ​​a target, it locates the target object based on the optimal docking point position and the visual data obtained by the depth camera, and adjusts the grasping position and posture in real time to ensure accurate grasping operation.

8. The method for continuous motion planning of a mobile working robot for multi-target grasping according to claim 7, characterized in that: When the mobile working robot performs multi-target grasping work, it obtains point cloud information of the laser radar in real time; when the laser radar detects an obstacle, it avoids the obstacle to ensure that the robot can safely and smoothly reach the next target area.

Images