Composite robot intelligent grasping planning method in high-hybrid warehouse environment

Abstract

The present application relates to a composite robot intelligent grabbing planning method in a high-hybrid warehouse environment, which converts multi-dimensional material distribution into a simplified and orderly structure by defining a virtual shelf position at the center line of the aisle; converts the composite robot access planning into a multi-objective optimization problem of the parking position; adopts a greedy strategy of the maximum covering circle model for work station basic planning, and then optimizes the work station through a covering optimization algorithm based on mutation; obtains the spatial relationship between the robot and the material, adopts a seventh-degree polynomial for grabbing trajectory planning, combines a time sequence distribution method, takes the weight sum of the total time, energy consumption and pulsating impact as the optimization target, and realizes comprehensive optimization of the trajectory under the premise of meeting the operation constraints. The present application can effectively simplify the material distribution in a complex warehouse environment, optimize the parking position and grabbing trajectory of the composite robot, improve the grabbing efficiency and reduce the energy consumption, and is suitable for intelligent grabbing operation in a high-hybrid warehouse scene.

Description

Technical Field

[0001] This invention belongs to the field of robot grasping trajectory planning technology, and particularly relates to a composite robot intelligent grasping planning method in a highly complex warehouse environment. Background Technology

[0002] The rapid development of highly flexible manufacturing has led to a continuous increase in demand for logistics and transportation. To cope with the increasingly complex needs of material storage management, highly complex and dynamic intelligent warehousing systems have gradually become an important development direction for the warehousing and logistics transportation industry. However, the rudimentary intelligent warehousing systems, which only demonstrate intelligence in non-execution aspects such as information management, scheduling, and sorting, are increasingly unable to meet the execution requirements of highly complex and flexible warehousing environments.

[0003] With the widespread application of hybrid robots integrating traditional serial robotic arms and automated guided vehicles (AGVs) in material handling tasks, material storage and retrieval systems are gradually developing towards higher flexibility and adaptability. Currently, research on hybrid robots in warehousing and industrial scenarios mainly focuses on path planning and control. Some researchers have proposed a collaborative optimization framework for path planning and task allocation to achieve orderly collaboration among different types of hybrid robots. Others have proposed a distributed multi-stage optimization method that integrates task decomposition, allocation, and scheduling strategies to solve the problem of cross-time scheduling dependencies for complex tasks of hybrid robots. Still others have proposed using wheeled mobile robots and robotic arms to collaboratively control electromechanical production lines, achieving efficient workpiece transfer and providing a reference solution for the industrial application of hybrid robots. Analysis reveals that existing hybrid robot planning methods generally assume that a single workstation corresponds to only one task, making it difficult to handle multiple mapping relationships as the number of materials increases. When facing highly mixed warehousing scenarios such as 3C components, the uncertainty of material distribution significantly increases the complexity of workstation tasks, and traditional single mapping modes cannot effectively meet the planning needs of complex workstations.

[0004] To improve the intelligence level of the composite robot's storage and retrieval strategy, this invention aims at composite robot material grasping in a highly mixed warehousing environment. Combining the Virtual Shelf-based Material Partitioning Scheme (MPVS) and the Mutation-based Coverage Optimization Algorithm (MBCO), it proposes an intelligent grasping strategy and a grasping trajectory planning method for composite robots. Summary of the Invention

[0005] The purpose of this invention is to provide a composite robot intelligent grasping planning method in a highly complex warehousing environment, so as to improve the intelligence level of composite robot storage and retrieval strategies in a highly complex and flexible warehousing environment.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] A method for intelligent grasping planning of composite robots in highly complex warehousing environments includes the following steps:

[0008] S1: By defining the location of virtual shelves along the aisle centerline, the multi-dimensional material distribution is transformed into a simplified and orderly structure;

[0009] S2: Perform workstation planning for the composite robot, transforming the composite robot access planning into a multi-objective optimization problem for docking positions;

[0010] S3: Achieve multi-objective optimization through a greedy strategy and a mutation coverage optimization algorithm using the maximum coverage circle model;

[0011] S4: Obtain the spatial relationship between the robot's docking station and the robotic arm's material space, use a seventh-order polynomial for grasping trajectory planning, and design a time series allocation method to optimize running time and energy consumption impact.

[0012] Preferably, the specific process of step S1 is as follows:

[0013] S11: Take the center axis of the aisle as the Y-direction reference, define several virtual shelf location points in this direction, corresponding to the actual shelf entities, and generate virtual shelves projected onto the center line of the aisle on both sides of the aisle.

[0014] S12: Project the coordinates of the material locations on the real shelves onto the corresponding virtual shelves in the aisles. The composite robot completes the material transfer between the left and right shelves simultaneously by rotating its joints.

[0015] Preferably, the specific process of step S3 is as follows:

[0016] S31: Basic planning of the workstations for composite robots based on a greedy strategy;

[0017] S32: The basic planning of the workstation of the composite robot is optimized by a mutation-based coverage optimization algorithm.

[0018] Preferably, the specific process of performing basic planning of the workstation of the composite robot based on a greedy strategy in step S31 is as follows:

[0019] S311: Define the center of the mapping circle on the virtual shelf for the robotic arm's specified sectional workspace when the composite robot is in different positions as O(x). o y o , z o ), where x o and y o Represents the X and Y coordinates within the robotic arm's space, z o The Z-axis coordinate represents the fitted circle in the workspace of the specified sectional surface of the robotic arm;

[0020] S312: Define the center point o of the robotic arm base a Coordinates are (x oa ,y oa ,z oa The center of the workspace fitting circle is at a height of h above the ground, therefore the coordinates of the center of the workspace fitting circle are (x...). oa y oa , h);

[0021] S313: For any stop point S j Set material p i The conditions for being covered by the robotic arm's workspace are:

[0022]

[0023] S314: Select the composite robot docking point that can currently cover the most uncovered materials. The specific selection process is as follows:

[0024]

[0025] Among them, S j * For the best stopping point, S candidate U represents the set of all possible docking points for composite robots in the current aisle, and U represents the set of materials that are currently uncovered.

[0026] S315: Remove the covered material from the uncovered set U:

[0027]

[0028] S316: Calculate coverage and docking points using the following formula:

[0029]

[0030] k = |S|;

[0031]

[0032] Wherein, C(s) j ) indicates stop point S j The collection of covered materials.

[0033] Preferably, the specific process of optimizing the basic planning of the workstation of the composite robot in step S32 using a mutation-based coverage optimization algorithm is as follows:

[0034] S321: By running a greedy strategy multiple times to generate multiple different initial solutions, set the set of materials to be taken as U = {u1, u2, ..., u...}. m}, where each u iThe candidate workstation set C = {c1, c2, ..., c3} represents the location of a material to be picked up. n}, where each c j S represents the position of a composite robot workstation. Each solution is represented as a set of candidate workstation index values. j ={c j1 ,c j2 ,…,c jk If the population size is set to N, then the initial solution set is: S = {S1, S2, ..., S} N};

[0035] The selection of candidate workstations employs a random strategy, with the specific formula as follows:

[0036]

[0037] in, C is the optimal docking point, C is the set of all possible docking points of the composite robot in the current aisle, U is the set of currently uncovered materials, α is the random weight, and κ is a random variable uniformly distributed in the interval (0, 1).

[0038] S322: By continuously selecting workstations, the new workstation is added to the workstation solution set S after the selection is completed. j The set of uncovered materials U is continuously updated until all materials are covered. The updated formula is as follows:

[0039]

[0040] S323: Define the fitness function f(S i Minimize the number of selected workstations for each solution S. i The fitness function f(S) is used to evaluate the merits of the fitness function. i ) is represented as:

[0041]

[0042] S324: Set the mutation probability to p m For each solution S i We set conditions for mutation to determine whether the solution has mutated. The conditions for mutation are as follows:

[0043] r <p m ;

[0044] Where r is a random number between 0 and 1;

[0045] If the solution is S i If a mutation occurs, k workstations are randomly selected from S. iRemove from the middle, and the solution set S of the workstation after removal is obtained. i ′ The expressions for the set of materials to be retrieved and the number of workstations to be removed, k, are as follows:

[0046]

[0047] S325: After removing the workstation, cover the uncovered materials again, and then check the current set of materials to be picked up, U. ′ Perform material coverage and add the new workstation to the workstation set S. i ′ In the middle, if f(S) is satisfied i ′ ) <f(S i If the solution is S, then update the solution by letting S. i =S i ′ .

[0048] Preferably, the specific process of step S4 is as follows:

[0049] S41: The joint jerk is constrained by a seventh-order polynomial, and the angular velocity, angular acceleration, and angular jerk time expressions of the robotic arm are obtained by taking multiple derivatives of the seventh-order polynomial, as detailed below:

[0050]

[0051] S42: By specifying the angle, velocity, acceleration value, and jerk of the trajectory's starting and ending points, and substituting the eight constraint equations into the expression, a system of constraint equations is obtained, as follows:

[0052]

[0053] S43: Optimize the time series allocation for the grasping trajectory planning.

[0054] Preferably, the specific process of optimizing the time series allocation of the grasping trajectory planning in step S43 is as follows:

[0055] S431: Using the weighted sum of the three parameters—total time value S1, energy consumption S2, and pulsating impact value S3—as the optimization objective equation, a set of balanced solutions for the trajectory that comprehensively considers the four characteristics is obtained. The optimization objective equation is:

[0056]

[0057] E = m1*S1 + m2*S2 + m3*S3;

[0058] Where S1 represents the total time the robotic arm spends on the trajectory; t i ti+1 S2 represents the time node value of adjacent trajectory points; S2 is used to measure the energy consumption of the robotic arm during trajectory operation. i S3 represents the angular acceleration of the i-th joint; S3 is used to measure the pulsating impact value of the robotic arm during trajectory operation, j i Represents the angular acceleration of the i-th joint; E is the optimization target value; m i The corresponding weight coefficients for the sub-optimization objectives are used to adjust the proportion of each optimization objective by adjusting the weight coefficients.

[0059] S432: Based on the operational constraints of the composite robot arm in terms of angle, velocity, acceleration, and jerk, the constraint conditions are set as follows:

[0060] |q(t)|≤q max |v(t)|≤v max |a(t)|≤a max |j(t)|≤j max ;

[0061] Where q represents the joint angle value, v represents the joint angular velocity value, a represents the joint angular acceleration value, and j represents the joint angular jerk value.

[0062] The beneficial effects of this invention include:

[0063] The present invention provides a composite robot intelligent grasping and planning method for highly complex warehousing environments. Firstly, by defining virtual shelves along the aisle centerline and projecting the actual material locations onto these virtual shelves, the multi-dimensional and disordered material distribution in highly complex warehousing is transformed into an ordered structure. This significantly reduces the interference of environmental complexity on planning, enabling the composite robot to simultaneously handle materials on both sides of the shelves through joint rotation. This provides a simple and unified spatial benchmark for subsequent grasping planning, improving overall planning efficiency.

[0064] Secondly, the planning of the composite robot's access is transformed into a multi-objective optimization problem of docking positions, and the greedy strategy of the maximum coverage circle model and the mutation coverage optimization algorithm are combined: the greedy strategy quickly realizes the basic workstation planning by prioritizing the docking point that covers the most uncovered materials, ensuring the initial coverage rate is maximized; the mutation optimization algorithm further reduces the number of required docking points through operations such as generating multiple initial solutions and randomly mutating and adjusting workstations, thereby reducing the number of robot movements and path length while ensuring high coverage, thus reducing energy consumption and time costs.

[0065] Furthermore, a seventh-order polynomial is used for grasping trajectory planning. By constraining parameters such as joint jerk, angular velocity, and angular acceleration, the smoothness of the trajectory is ensured, reducing vibration and impact during the operation of the robotic arm. Combined with a time series allocation method, the total time, energy consumption, and pulsating impact are used as the optimization target. Under the premise of satisfying the physical constraints of the robotic arm, a comprehensive balance between efficiency, energy consumption, and equipment wear is achieved, improving the stability and economy of grasping operations.

[0066] Finally, the overall design is tailored for highly complex warehousing environments. By simplifying material distribution through virtual shelving, reducing the number of workstations through optimized algorithms, and reducing equipment load through smooth trajectories, it effectively addresses the needs of complex scenarios such as dense materials, disorderly distribution, and multi-shelf operations, thereby improving the adaptability and practicality of the composite robot in actual warehousing operations. Attached Figure Description

[0067] Figure 1 This is a top view of the virtual shelf of the present invention.

[0068] Figure 2 This is a three-dimensional perspective schematic diagram of the virtual shelf of the present invention.

[0069] Figure 3 This is a diagram illustrating the planned material distribution 3 in scenario 3 of the present invention.

[0070] Figure 4 This is a schematic diagram showing the coordinate relationship between the various positions of the composite robot and the material to be picked up according to the present invention.

[0071] Figure 5 This is the trajectory diagram for the polynomial trajectory planning of the dual material transfer task of the present invention.

[0072] Figure 6 This is a schematic diagram of the motion characteristics of the end effector of the present invention.

[0073] Figure 7 A comparison chart of the maximum jump values ​​of the trajectory planning method of the present invention.

[0074] Figure 8 This is a speed diagram of the end effector of the robotic arm according to the present invention.

[0075] Figure 9 This is a schematic diagram of the intelligent grasping and planning method for composite robots in a highly complex warehousing environment according to the present invention. Detailed Implementation

[0076] The following is in conjunction with the appendix Figures 1-9 The present invention will be further described in detail below:

[0077] See appendix Figure 9As shown, a composite robot intelligent grasping planning method for high-complexity warehouse environments is characterized by the following steps:

[0078] S1: By defining the location of virtual shelves along the aisle centerline, the multi-dimensional material distribution is transformed into a simplified and orderly structure;

[0079] S2: Perform workstation planning for the composite robot, transforming the composite robot access planning into a multi-objective optimization problem for docking positions;

[0080] S3: Achieve multi-objective optimization through a greedy strategy and a mutation coverage optimization algorithm using the maximum coverage circle model;

[0081] S4: Obtain the spatial relationship between the robot's docking station and the robotic arm's material space, use a seventh-order polynomial for grasping trajectory planning, and design a time series allocation method to optimize running time and energy consumption impact.

[0082] In this embodiment, an intelligent material partitioning scheme is first proposed for the partitioning and material retrieval planning problem in highly mixed warehousing environments. By defining virtual shelf positions along the aisle centerline, the multi-dimensional material distribution is transformed into a simpler and more ordered structure. Then, the composite robot storage and retrieval planning is transformed into a multi-objective optimization problem of docking positions. A greedy strategy based on the maximum coverage circle model and a variant coverage optimization algorithm (MBCO) are proposed, and the algorithm's advancement is verified through multi-scenario experiments. Finally, the spatial relationship between the robot's docking station and the robotic arm's material is analyzed. A seventh-order polynomial is used for trajectory planning, and a time-series allocation method is designed to optimize running time and energy consumption impact, thereby improving the overall planning effect.

[0083] The specific process of step S1 is as follows:

[0084] S11: Take the center axis of the aisle as the Y-direction reference, define several virtual shelf location points in this direction, corresponding to the actual shelf entities, and generate virtual shelves on both sides of the aisle centerline that are projected onto the aisle centerline. The virtual shelves do not have a real physical structure, but are used as an abstract data structure in the algorithm.

[0085] S12: Project the coordinates of the material locations on the real shelves onto the corresponding virtual shelves in the aisles. The composite robot completes the material transfer between the left and right shelves simultaneously by rotating its joints.

[0086] In warehouse layout, materials are discretely distributed at different heights and distances on shelves according to different storage rules. In order to more effectively divide and plan the material storage and retrieval of composite robots, this invention proposes a material partitioning scheme based on virtual shelving (MPVS). By defining the position of virtual shelving along the center line of the aisle, the multi-dimensional material distribution is transformed into a more simplified and orderly structure.

[0087] When locating and constructing virtual shelving, the center axis of the aisle is taken as the Y-direction reference. Several virtual shelving position points are defined along this direction, corresponding to the actual shelving entities. Virtual shelving is generated along the center line of each aisle, projecting the real shelving onto the aisle center line. The virtual shelving does not have a real physical structure but serves as an abstract data structure in the algorithm. Simultaneously, the coordinates of the material positions distributed on the real shelving are projected onto the corresponding virtual shelving in the aisle. The composite robot can simultaneously transfer materials between the left and right shelving via joint rotation. A top view of the virtual shelving is shown below. Figure 1 As shown in the diagram, the stereoscopic view is as follows: Figure 2 As shown.

[0088] In a warehouse environment, materials are distributed across multiple shelves, and composite robots need to transfer these materials at specific docking points. For materials stored in different locations on different shelves... i (x i y i , z i ), Generate a set of materials to be picked up, P = {p1, p2, ..., p ... i}, plan the set of docking points for the composite robot S = {s1, s2, ..., s} j}, where each stop s j The position is (x j ,y j ), x j y represents the current location of the composite robot in the aisle. j Indicates the docking position on the Y-axis.

[0089] When planning workstations for composite robots, the robot docking station planning problem is transformed into a workspace coverage problem on a specified cross-section. Using a greedy strategy as the basic idea, the storage area is decomposed into several independent aisle units. Within the aisle range, with the set of materials to be retrieved as input, the robot progressively selects docking points that can cover the most remaining materials. After selecting a docking point, priority is given to choosing the docking point that can cover the most uncovered materials, and the materials covered by that point are removed from the list of materials to be retrieved. This selection process is repeated until all materials to be retrieved are covered.

[0090] The specific process of step S3 is as follows:

[0091] S31: Basic planning of the workstations for composite robots based on a greedy strategy;

[0092] S32: The basic planning of the workstation of the composite robot is optimized by a mutation-based coverage optimization algorithm.

[0093] The specific process of performing basic planning for the workstation of the composite robot based on a greedy strategy in step S31 is as follows:

[0094] S311: Define the center of the mapping circle on the virtual shelf for the robotic arm's specified sectional workspace when the composite robot is in different positions as O(x). O ,y O ,z O ), where x O and y O Represents the X and Y coordinates within the robotic arm's space, z O The Z-axis coordinate represents the fitted circle in the workspace of the specified sectional surface of the robotic arm;

[0095] S312: Define the center point o of the robotic arm base a Coordinates are (x oa ,y oa ,z oa The center of the workspace fitting circle is at a height of h above the ground, therefore the coordinates of the center of the workspace fitting circle are (x...). oa y oa ,h);

[0096] S313: For any stop point S j Set material p i The conditions for being covered by the robotic arm's workspace are:

[0097]

[0098] S314: Select the composite robot docking point that can currently cover the most uncovered materials. The specific selection process is as follows:

[0099]

[0100] Among them, S j * For the best stopping point, S candidate U represents the set of all possible docking points for composite robots in the current aisle, and U represents the set of materials that are currently uncovered.

[0101] S315: Remove the covered material from the uncovered set U:

[0102]

[0103] S316: Calculate coverage and docking points using the following formula:

[0104]

[0105] k = |S|;

[0106]

[0107] Wherein, C(s) j ) indicates stop point S j The collection of covered materials.

[0108] Due to the greater complexity and more dispersed distribution of materials in actual warehousing environments, traditional greedy strategies are prone to getting trapped in local optima and failing to find a globally optimal solution. Therefore, this invention proposes a Mutation-Based Coverage Optimization (MBCO) algorithm, aiming to provide a more efficient, simpler, and better suited optimization method for warehouse workstation planning scenarios. The MBCO algorithm gradually optimizes coverage by randomly removing and adding candidate workstations. In the initialization phase, the algorithm uniformly generates candidate workstations in each aisle and uses the greedy strategy solution as the initial solution for the MBCO algorithm. Subsequently, through multiple iterations, the algorithm performs a mutation operation on each solution, randomly removing some workstations and adding new ones to explore better solutions.

[0109] The specific process of optimizing the basic planning of the workstation of the composite robot in step S32 using a mutation-based coverage optimization algorithm is as follows:

[0110] S321: By generating multiple different initial solutions through multiple runs of a greedy strategy, set the set of materials to be taken as U = {u1, u2, ..., u...}. m}, where each u i The candidate workstation set C = {c1, c2, ..., c3} represents the location of a material to be picked up. n}, where each c j S represents the position of a composite robot workstation. Each solution is represented as a set of candidate workstation index values. j ={c j1 ,c j2 ,…,c jk Let the population size be N, then the initial solution set is: S = {S1, S2, ..., S...} N};

[0111] The selection of candidate workstations employs a random strategy, with the specific formula as follows:

[0112]

[0113] in, C is the optimal docking point, U is the set of all possible docking points of the composite robot in the current aisle, α is the set of uncovered materials, k is a random weight, and k is a random variable uniformly distributed in the interval (0,1) to break the balance deadlock when the same amount of coverage is achieved.

[0114] S322: By continuously selecting workstations, the new workstation is added to the workstation solution set S after the selection is completed. j The set of uncovered materials U is continuously updated until all materials are covered. The updated formula is as follows:

[0115]

[0116] S323: Define the fitness function f(S i Minimize the number of selected workstations for each solution S. i The fitness function f(S) is used to evaluate the merits of the fitness function. i ) is represented as:

[0117]

[0118] S324: Set the mutation probability to p m For each solution S i We set conditions for mutation to determine whether the solution has mutated. The conditions for mutation are as follows:

[0119] r <p m ;

[0120] Where r is a random number between 0 and 1;

[0121] If the solution is S i If a mutation occurs, k workstations are randomly selected from S. i Remove from the middle, and the solution set S of the workstation after removal is obtained. i ′ The expressions for the set of materials to be retrieved and the number of workstations to be removed, k, are as follows:

[0122]

[0123] S325: After removing the workstation, cover the uncovered materials again, and then check the current set of materials to be picked up, U. ′ Perform material coverage and add the new workstation to the workstation set S. i ′ In the middle, if f(S) is satisfied i ′ ) <f(S i If the solution is S, then update the solution by letting S. i =Si ′ .

[0124] To verify the superiority of the MBCO algorithm in the planning of workstations for composite robots, this invention designs a simulation comparison experiment. The traditional greedy strategy, EG-ACO algorithm, EG-SSA algorithm, and MBCO algorithm are used as comparison groups. Three warehouse scenarios are set up, with the following specifications: shelf width 0.4m, shelf length 7m, shelf height 1.4m, minimum shelf storage height 0.4m, and aisle width 1.2m. The number of shelf layers and the maximum number of materials that can be stored per layer are set differently. Specific parameters are shown in Table 1.

[0125] Table 1 Parameter Settings for Different Scenarios

[0126] Scene number Number of shelf layers Maximum number of materials per layer 1 3 15 2 5 21 3 5 30

[0127] For each scenario, three independent experiments were conducted, with different material distributions in each experiment to simulate the diversity of the actual production environment. First, the materials and their distributions for each experiment were randomly generated within the scenario to ensure representativeness and diversity. Second, greedy strategies, EG-ACO algorithms, EG-SSA algorithms, and MBCO algorithms were used for workstation planning under the same material distribution in the same scenario. Finally, the planning results, computation time, and number of planned workstations for each algorithm were recorded in each experiment. The planning results of the four algorithms under different scenarios and material distributions are shown in Tables 2, 3, and 4.

[0128] Table 2. Workstation Planning Effects under Different Material Distributions in Scenario 1

[0129]

[0130] Table 3. Workstation Planning Effects under Different Material Distributions in Scenario 2

[0131]

[0132] Table 4. Workstation Planning Effects under Different Material Distributions in Scenario 3

[0133]

[0134] To highlight the comparison results of different algorithms and intuitively demonstrate the planning effects of each algorithm, the planning effect diagrams for material distribution 3 in Scenario 3 are shown below. Figure 3 As shown:

[0135] As shown in the charts, in simple scenarios, all algorithms achieve 100% material coverage. The Greedy algorithm has an average of 6.33 workstations, while EG-ACO, EG-SSA, and MBCO all have 5.67, a reduction of approximately 14.1%, indicating a significant advantage in resource allocation for the improved solutions. Although Greedy has extremely short computation time, it has shortcomings when dealing with complex material distributions. In medium-complex scenarios, under material distribution 1, Greedy only achieves 85.71% coverage and plans 9 workstations, while EG-ACO and MBCO both achieve full coverage, reducing the number of workstations to 7, a reduction of approximately 22.2% compared to Greedy. MBCO has the lowest computation time, approximately 0.19 seconds. In highly complex scenarios, Greedy needs to plan 16 workstations under material distribution 3, while EG-ACO and MBCO both control it to 10, a reduction of 37.5% and 25% respectively compared to Greedy, and EG-SSA requires 12. In terms of computation time, MBCO consistently maintains the lowest latency, with its runtime latency reduced by nearly 80% compared to EG-ACO and slightly better than EG-SSA. In summary, the MBCO algorithm, while ensuring 100% coverage, demonstrates excellent resource utilization efficiency and high efficiency by reducing the number of workstations and shortening computation time, providing a reliable and economical technical solution for the optimized design of multi-layer racking and large-scale material storage systems.

[0136] In a warehousing environment, when a composite robot receives information about materials to be retrieved, it can determine the coordinates of the target work location through workstation planning and then reach the work location through specific spatial positioning, path planning, and motion control. When the composite robot is at different workstations, the coordinates between the center position of the robot's moving chassis, the position of the robotic arm base, the starting position of the robotic arm's end effector, the material placement point, and the corresponding material to be retrieved must satisfy… Figure 4 The spatial relationships shown.

[0137] Analysis of the material handling tasks of the composite robot at different docking positions reveals that the material handling tasks are mainly divided into three stages: material grasping, material placement, and returning to the starting point. The material grasping stage refers to the robotic arm's end effector moving from the starting point to the material point and pausing for a certain period to complete the material grasping. The material placement stage refers to the robotic arm's end effector moving from the material point to the placement point and pausing for a period to complete the placement. The returning to the starting point stage is to ensure the overall stability of the composite robot during movement; therefore, after each material handling task, the robotic arm's end effector moves to the starting point, where overall stability is high. Thus, the velocity, acceleration, and jerk of the starting and ending points of the segmented trajectory of the robotic arm in each stage are all zero, placing high demands on the constraints of joint variables. Therefore, a seventh-order polynomial is used for the robotic arm's joint space trajectory planning method.

[0138] The specific process is as follows:

[0139] S41: The joint jerk is constrained by a seventh-order polynomial, and the angular velocity, angular acceleration, and angular jerk time expressions of the robotic arm are obtained by taking multiple derivatives of the seventh-order polynomial, as detailed below:

[0140]

[0141] S42: By specifying the angle, velocity, acceleration value, and jerk of the trajectory's starting and ending points, and substituting the eight constraint equations into the expression, a system of constraint equations is obtained, as follows:

[0142]

[0143] S43: Optimize the time series allocation of the grasping trajectory planning to ensure that the trajectory of the robotic arm meets all constraints while reducing unnecessary energy consumption and trajectory pulsation impact, thereby improving the efficiency and stability of the system.

[0144] To ensure optimal time, energy, and impact for the composite robot during material handling tasks, this paper optimizes the time series allocation of the trajectory planning part. This ensures that the robot's trajectory meets all constraints while minimizing unnecessary energy consumption and trajectory pulsation impact, thereby improving the system's efficiency and stability. Based on an improved ant colony optimization algorithm, this invention adjusts the time series of each stage of the task through iterative optimization, achieving an optimal balance between total execution time, energy consumption, and trajectory pulsation impact. This ensures the smoothness and controllability of the robot during task execution.

[0145] The specific process of optimizing the time series allocation for the grasping trajectory planning in step S43 is as follows:

[0146] S431: Using the weighted sum of the three parameters—total time value S1, energy consumption S2, and pulsating impact value S3—as the optimization objective equation, a set of balanced solutions for the trajectory that comprehensively considers the four characteristics is obtained. The optimization objective equation is:

[0147]

[0148] E = m1*S1 + m2*S2 + m3*S3;

[0149] Where S1 represents the total time the robotic arm spends on the trajectory; t i t i+1 S2 represents the time node value of adjacent trajectory points; S2 is used to measure the energy consumption of the robotic arm during trajectory operation. i S3 represents the angular acceleration of the i-th joint; S3 is used to measure the pulsating impact value of the robotic arm during trajectory operation, j iRepresents the angular acceleration of the i-th joint; E is the optimization target value; m i The corresponding weight coefficients for the sub-optimization objectives are used to adjust the proportion of each optimization objective by adjusting the weight coefficients.

[0150] S432: Based on the operational constraints of the composite robot arm in terms of angle, velocity, acceleration, and jerk, the constraint conditions are set as follows:

[0151] |q(t)|≤q max |v(t)|≤v max |a(t)|≤a max |j(t)|≤j max ;

[0152] Where q represents the joint angle value, v represents the joint angular velocity value, a represents the joint angular acceleration value, and j represents the joint angular jerk value.

[0153] To demonstrate the rationality and superiority of the grasping scheme based on the improved 7th-order polynomial design, the same material grasping task under the same scenario and material distribution was selected as the planning object. The 3rd, 5th, and 7th-order polynomial interpolation trajectory planning methods were applied to the above planning object for trajectory planning. Similarly, the planning effect of a dual-material grasping task on the same side under a certain docking station was selected, and the trajectory dynamic characteristic curves under each method were plotted, as shown in the figure below. Figure 5 and Figure 6 As shown:

[0154] To quantitatively compare the results of different trajectory planning methods, this paper uses the abrupt change in the derivative of a physical quantity within a very short time as the jump value, and evaluates the merits of each planning method by combining the jump value with the dynamic characteristic curves of the robotic arm plotted by different planning methods. Comparison of maximum jump values ​​of different trajectory planning methods. Figure 7 .

[0155] It can be seen that the velocity, acceleration, and jerk jump values ​​planned by the cubic polynomial are 231, 12066.43, and 900548.69, respectively. Combined with the graph, it can be seen that the cubic polynomial interpolation trajectory planning method can only smoothly plan the velocity; the acceleration and jerk parameters have many abrupt changes. The velocity, acceleration, and jerk jump values ​​planned by the quintic polynomial are 20.55, 736.98, and 37772.21, respectively. The quintic polynomial has a certain advantage over the cubic polynomial, but it still cannot achieve smooth jerk, and the acceleration and jerk jump values ​​are both relatively large. Finally, the velocity, acceleration, and jerk values ​​calculated using the seventh-order polynomial were 15.28, 92.86, and 2718.03, respectively, representing reductions of 93.39%, 93.89%, and 99.70% compared to the cubic polynomial, and reductions of 25.64%, 87.40%, and 92.80% compared to the quintic polynomial, while maintaining the continuity and smoothness of the curves. To minimize impact during the robotic arm's movement, this paper employs seventh-order polynomial interpolation for trajectory planning.

[0156] See Figure 8 As shown, when performing material transfer tasks, the end effector's moving speed is maintained at around 0-0.3 m / s, which meets the requirements for the moving speed of the end effector of the robotic arm in the warehousing environment. The joint angle changes are continuous without abrupt changes or peaks. The acceleration and jerk images are smooth and continuous without sharp peak changes, indicating that the robotic arm can smoothly transition from one posture to another. At the same time, it ensures that there is no severe vibration or impact when the robotic arm runs along the expected trajectory. Therefore, the optimized trajectory can meet the actual material transfer operation requirements.

[0157] In summary, the intelligent grasping and planning method for composite robots in highly complex warehousing environments provided by this invention transforms the multi-dimensional and disordered material distribution in highly complex warehousing into an ordered structure by defining virtual shelves along the aisle centerline and projecting the real material positions onto the virtual shelves. This reduces the interference of environmental complexity on planning, enabling the composite robot to simultaneously handle materials on both sides of the shelves through joint rotation. The composite robot's storage and retrieval planning is transformed into a multi-objective optimization problem of docking positions, combining a greedy strategy based on the maximum coverage circle model with a mutated coverage optimization algorithm: the greedy strategy prioritizes docking points that cover the most uncovered materials, quickly achieving basic workstation planning; the mutated optimization algorithm further reduces the number of required docking points through multiple initial solution generation and random mutation adjustment of workstations, ensuring high coverage while reducing the number of robot movements and path length. A seventh-order polynomial is used for grasping trajectory planning. By constraining parameters such as joint accelerometer, angular velocity, and angular acceleration, vibration and impact during the operation of the robotic arm are reduced. Combined with a time series allocation method, the total time, energy consumption, and pulsating impact are used as the optimization target. Under the premise of satisfying the physical constraints of the robotic arm, a comprehensive balance between efficiency, energy consumption, and equipment wear is achieved.

Claims

1. A composite robot intelligent grasping planning method for high-complexity warehousing environments, characterized in that, Includes the following steps: S1: By defining the location of virtual shelves along the aisle centerline, the multi-dimensional material distribution is transformed into a simplified and orderly structure; S2: Perform workstation planning for the composite robot, transforming the composite robot access planning into a multi-objective optimization problem for docking positions; S3: Achieve multi-objective optimization through a greedy strategy and a mutation coverage optimization algorithm using the maximum coverage circle model; S4: Obtain the spatial relationship between the robot's docking station and the robotic arm's material space, use a seventh-order polynomial for grasping trajectory planning, and design a time series allocation method to optimize running time and energy consumption impact; The specific process of step S3 is as follows: S31: Basic planning of the workstations for composite robots based on a greedy strategy; S32: The basic planning of the workstation for the composite robot is optimized using a mutation-based coverage optimization algorithm; The specific process of performing basic planning for the workstation of the composite robot based on a greedy strategy in step S31 is as follows: S311: Set the center of the mapping circle on the virtual shelf for the robotic arm's specified sectional workspace when the composite robot is in different positions. in, and Represents the X and Y coordinates within the robotic arm's space. The Z-axis coordinate represents the fitted circle in the workspace of the specified sectional surface of the robotic arm; S312: Define the center point of the robotic arm base Coordinates are The height of the center of the fitted circle in the workspace above the ground is h Therefore, the coordinates of the center of the fitted circle in the workspace are... ; S313: For any stop point Set materials The conditions for being covered by the robotic arm's workspace are: ； S314: Select the composite robot docking point that can currently cover the most uncovered materials. The specific selection process is as follows: ； in, For the best stopping point, This is the set of all possible docking points for the composite robot in the current aisle. U This is the set of materials that are currently not covered. S315: The material to be covered has never been covered. U Removed from: ； S316: Calculate coverage and docking points using the following formula: ； ； ； in, Indicates the stop point The collection of covered materials.

2. The intelligent grasping and planning method for composite robots in a highly complex warehousing environment according to claim 1, characterized in that, The specific process of step S1 is as follows: S11: Take the center axis of the aisle as the Y-direction reference, define several virtual shelf location points in the Y-direction, corresponding to the actual shelf entities, and generate virtual shelves projected onto the center line of the aisle on both sides of the aisle. S12: Project the coordinates of the material locations on the real shelves onto the corresponding virtual shelves in the aisles. The composite robot completes the material transfer between the left and right shelves simultaneously by rotating its joints.

3. The intelligent grasping and planning method for composite robots in a highly complex warehousing environment according to claim 1, characterized in that, The specific process of optimizing the basic planning of the workstation of the composite robot in step S32 using a mutation-based coverage optimization algorithm is as follows: S321: Generate multiple different initial solutions by running the greedy strategy multiple times, and set the set of materials to be taken. , each of which Represents the location of a material to be picked up; a set of candidate workstations. Each of them Representing a composite robot workstation position, each solution is represented as a set of candidate workstation index values, i.e. Set the population size to N Then the initial solution set is: ; The selection of candidate workstations employs a random strategy, with the specific formula as follows: ； in, For the best stopping point, C This is the set of all possible docking points for the composite robot in the current aisle. U This is the set of materials that are currently not covered. For random weights, It is a random variable that is uniformly distributed in the interval (0, 1); S322: By continuously selecting workstations, the new workstation is added to the workstation solution set after the selection is completed. The set of uncovered materials is continuously updated. U until all materials are covered, i.e. The updated formula is as follows: ； ； S323: Define the fitness function Minimize the number of selected workstations for each solution The fitness function is used to evaluate the merits of each. Represented as: ； S324: Set the mutation probability to... For each solution We set conditions for mutation to determine whether the solution has mutated. The conditions for mutation are as follows: ； in, A random number between 0 and 1; If the solution If a mutation occurs, then random selection will be made. k Each workstation Remove from the middle, and the solution set of the removed workstations. Collection of materials to be picked up and removal station k The expression is as follows: ； ； ； S325: After removing the workstation, cover the uncovered materials again, and then check the current set of materials to be picked up. Perform material coverage and add the new workstation to the workstation collection. In the middle, if the following conditions are met Then update the solution, let .

4. The intelligent grasping and planning method for composite robots in a highly complex warehousing environment according to claim 3, characterized in that, The specific process of step S4 is as follows: S41: The joint jerk is constrained by a seventh-order polynomial, and the angular velocity, angular acceleration, and angular jerk time expressions of the robotic arm are obtained by taking multiple derivatives of the seventh-order polynomial, as detailed below: ； S42: By specifying the angle, velocity, acceleration value, and jerk of the trajectory's starting and ending points, and substituting the eight constraint equations into the expression, a system of constraint equations is obtained, as follows: ； S43: Optimize the time series allocation for the grasping trajectory planning.

5. The intelligent grasping and planning method for composite robots in a highly complex warehousing environment according to claim 4, characterized in that, The specific process of optimizing the time series allocation for the grasping trajectory planning in step S43 is as follows: S431: Using the weighted sum of the three parameters—total time value S1, energy consumption S2, and pulsating impact value S3—as the optimization objective equation, a set of balanced solutions for the trajectory that comprehensively considers the four characteristics is obtained. The optimization objective equation is: ； ； ； ； Where S1 represents the total time the robotic arm spends on the trajectory; , S2 represents the time node value of adjacent trajectory points; S2 is used to measure the energy consumption of the robotic arm during trajectory operation. Indicates the first i The angular acceleration of each joint; S3 is used to measure the pulsating impact value of the robotic arm during trajectory operation. Indicates the first i The angle of each joint is accelerated; E To optimize the target value; The corresponding weight coefficients for the sub-optimization objectives are used to adjust the proportion of each optimization objective by adjusting the weight coefficients. S432: Based on the operational constraints of the composite robot arm in terms of angle, velocity, acceleration, and jerk, the constraint conditions are set as follows: ； in, q Indicates the joint angle value. v This represents the joint angular velocity value. a This represents the joint angular acceleration value. j This represents the joint angle acceleration value.

Images