Chess collecting method based on multi-piece cooperative grabbing of chess interaction device

Abstract

The application discloses a chess collecting method based on multi-chess cooperative grabbing of a chess interaction device, and belongs to the technical field of robot control. The method acquires the positions, postures, types and adjacent relations of all chesses on a chessboard through a vision unit and constructs a chess state atlas. The chesses on the chessboard are dynamically clustered and grouped into multiple cooperative grabbing groups by taking minimization of grabbing cost and probability of end effector interference being not more than a preset threshold as constraints. A multi-end effector cooperative grabbing instruction limiting grabbing sequence, grabbing path and grabbing force parameters is generated for each cooperative grabbing group. The multiple end effectors are controlled to perform grabbing on the chesses in the same cooperative grabbing group in parallel and transfer the chesses to a chess collecting area in the same time window. The method improves the grouping rationality and conflict-free scheduling capability of multi-chess parallel grabbing, and solves the problems of low efficiency, poor adaptability and high motion interference risk of the existing chess collecting method in the scene of unstructured distribution of chesses.

Description

Technical Field

[0001] This invention relates to the field of robotic arm control technology, and in particular to a method for collecting chess pieces using a chess-playing interactive device based on multi-piece collaborative grasping. Background Technology

[0002] Interactive chess-playing devices are widely used in teaching, training, and entertainment scenarios for chess, Go, and Chinese chess. Automatic piece collection is a crucial element in improving the ease of use and automation of these devices. Currently, piece collection in interactive chess-playing devices typically relies on manual picking up of pieces, or a single end effector built into the device individually grabbing the pieces and moving them to the collection area. Some devices use suction cup arrays or magnetic components to collect pieces individually or in fixed rows. While these methods are simple in structure and easy to implement, the end effector needs to repeatedly travel between the chessboard and the collection area during the collection process. This high frequency of movement and long travel distance result in a long collection time, especially noticeable in chess games with a large number of pieces.

[0003] To improve game completion efficiency, existing technologies have attempted to use multiple end effectors to grasp multiple pieces in parallel. However, current multi-end effector-based game completion methods typically only allocate grasping tasks according to preset partitions or fixed intervals, without global planning based on the actual distribution of pieces on the board. After the game ends, the positions of pieces on the board are often randomly scattered, with local areas potentially showing dense, stacked, or clustered pieces. Fixed partitions or preset grasping strategies are ill-suited to this unstructured distribution, easily leading to overlapping movement spaces and path interference between end effectors, or unreasonable grouping resulting in some end effectors being idle and the overall game completion time not being significantly reduced. Furthermore, existing methods fail to comprehensively consider factors such as piece spacing, piece type differences, and the probability of overlapping movement spaces when generating grasping instructions for multiple end effectors, lacking quantitative optimization of grasping costs and interference risks, thus limiting the stability and success rate of parallel grasping by multiple end effectors.

[0004] Furthermore, existing technologies have not effectively utilized the position, posture, type, and adjacency relationships of pieces on the chessboard to construct a global state description, nor have they dynamically grouped and generated collaborative grasping instructions based on this global state. When the distribution of pieces is complex or there are locally dense areas, fixed grouping methods cannot guarantee that all pieces within each group can be reliably grasped by the corresponding end effector, and there is a lack of a unified conflict-free scheduling mechanism between different groups, which easily leads to mechanical interference or grasping failures. Therefore, existing chess-playing interactive device methods suffer from low grasping efficiency, poor grouping rationality, high risk of multi-end interference, and insufficient adaptability in scenarios where multiple pieces are grasped in parallel. It is necessary to provide a chess-playing method that can perform multi-piece collaborative grouping and conflict-free grasping scheduling based on the real-time state of the chessboard. Summary of the Invention

[0005] The purpose of this invention is to provide a chess-playing interactive device method based on multi-piece collaborative grasping. By constructing a chess piece state graph and dynamically clustering and grouping randomly distributed chess pieces on the chessboard based on the optimization objectives of minimizing the grasping cost and minimizing the interference probability of the end effectors, a collaborative grasping instruction for multiple end effectors is generated. This improves the rationality of grouping and the conflict-free scheduling capability of multiple end effectors grasping chess pieces in parallel, and solves the problems of low chess-playing efficiency, poor grouping adaptability, and high risk of motion interference of multiple end effectors in existing chess-playing methods in scenarios with unstructured chess piece distribution.

[0006] To address the aforementioned technical problems, a first aspect of this invention provides a method for collecting chess pieces using a chess-playing interactive device based on multi-piece collaborative grasping. The chess-playing interactive device includes a vision unit, a control unit, and multiple end effectors. The vision unit is used to acquire a chessboard image, the control unit is used to generate collaborative grasping instructions, and the multiple end effectors are used to grasp chess pieces. The method includes the following steps: S1: The visual unit is used to obtain the position, posture, type and adjacency relationship between all pieces on the chessboard, and a piece state graph is constructed based on the position, posture, type and adjacency relationship. The adjacency relationship is the relationship that the minimum distance between two pieces does not exceed a preset adjacency distance threshold. The piece state graph is used to characterize the spatial position relationship and state association between each piece on the chessboard and its adjacent pieces. S2: Based on the chess piece state graph, the chess pieces on the chessboard are dynamically clustered and grouped to form multiple collaborative grasping groups; when performing the dynamic clustering and grouping, the optimization objective is to minimize the grasping cost under the constraint that the interference probability of the end effector does not exceed a preset interference threshold. The end effector is the execution component in the chess interaction device used to grasp chess pieces. The grasping cost is the grasping cost value determined by the chess piece spacing and chess piece type. The interference probability is the probability that the movement spaces of multiple end effectors overlap. The preset interference threshold is the upper limit allowed for the probability that the movement spaces of end effectors overlap. The collaborative grasping group is a set of chess pieces consisting of two or more chess pieces that can be grasped in parallel by multiple end effectors in the same time window. S3: For each of the cooperative grasping groups, generate a multi-end-efficiency actuator cooperative grasping instruction that matches the number and spatial distribution of pieces within the cooperative grasping group. The multi-end-efficiency actuator cooperative grasping instruction defines the grasping order, grasping path, and grasping force parameters of each of the multiple end-efficiencies for the corresponding pieces. S4: According to the multi-end-efficiency actuator collaborative grasping instruction, control the multiple end-efficiency actuators to simultaneously perform grasping operations on the pieces in the same collaborative grasping group, and transfer the grasped pieces to the collection area until all pieces on the board have been transferred.

[0007] Furthermore, step S2, which involves dynamically clustering and grouping the pieces on the chessboard according to the piece state graph to form multiple collaborative fetching groups, includes: S21: Take all the pieces in the chess piece state map as the objects to be grouped, and according to the adjacency relationship and spacing between the pieces in the chess piece state map, aggregate the pieces that meet the preset proximity conditions in space to form multiple initial groups. The preset proximity condition is that the minimum spacing between two pieces does not exceed the maximum opening and closing distance of the end effector. S22: For each initial group, determine whether the initial group satisfies the condition that the same group can be grabbed in parallel. The condition that the same group can be grabbed in parallel is that the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, and the area enclosed by the smallest bounding polygon formed by all the pieces in the initial group on the chessboard plane does not contain other pieces not assigned to the initial group. S23: Perform a boundary adjustment operation on the initial group that does not meet the condition that the same group can be grabbed in parallel. The boundary adjustment operation is to remove the pieces that cause the area enclosed by the minimum bounding polygon to contain other pieces from the initial group, or to merge the pieces that are spatially continuous with the initial group and can be grabbed by the same end effector into the initial group. The compatible grabbing means that the piece type belongs to the set of piece types that can be processed by the same end effector. S24: For all groups formed after the boundary adjustment operation, calculate the end effector interference probability between any two groups. For two groups whose end effector interference probability exceeds a preset interference threshold, first redistribute the pieces in the two groups. If the end effector interference probability between the two groups still exceeds the preset interference threshold after redistribution, then merge the two groups. After each operation, recalculate the end effector interference probability between all pairs of groups until the end effector interference probability between all pairs of groups does not exceed the preset interference threshold. Output the final group as multiple collaborative grabbing groups.

[0008] Further, the step S22, determining whether the initial group satisfies the condition that the same group can be crawled in parallel, includes: S221: For each initial group, obtain the position coordinates of all pieces in the initial group on the chessboard plane, and determine the minimum bounding polygon of all pieces in the initial group. The area enclosed by the minimum bounding polygon is referred to as the spatial envelope. S222: For each other piece outside the initial group that is not assigned to the initial group, determine whether the position coordinates of the unassigned piece are located in the inner region of the spatial envelope, or whether the minimum distance between the position coordinates of the unassigned piece and the boundary of the spatial envelope is less than a preset envelope boundary distance threshold, wherein the preset envelope boundary distance threshold is determined based on the end effector end radius; S223: If the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, and the spatial envelope does not contain any other pieces not assigned to the initial group, and the minimum distance between the boundary of the spatial envelope and any unassigned piece is not less than the preset envelope boundary distance threshold, then the initial group is determined to meet the condition for parallel capture within the same group; otherwise, the initial group is determined not to meet the condition for parallel capture within the same group.

[0009] Further, the boundary adjustment operation performed in step S23 on the initial groups that do not meet the conditions for parallel crawling within the same group includes: S231: For an initial group that does not meet the condition that the same group can be grabbed in parallel, determine that the area enclosed by the minimum bounding polygon contains unassigned pieces within the group, and remove the pieces within the group from the initial group. S232: For unassigned pieces that are spatially continuous with the initial group and can be captured by the same end effector, calculate the minimum distance between the unassigned pieces and the remaining pieces in the initial group. If the minimum distance does not exceed the maximum opening and closing distance of the end effector, then the unassigned pieces are incorporated into the initial group. S233: Repeat S231 and S232 until the initial group meets the condition that the same group can be crawled in parallel.

[0010] Further, the calculation of the end effector interference probability between any two groups in step S24 includes: S241: For any two groups, obtain the three-dimensional motion trajectory of the end effector corresponding to all pieces in the two groups within a preset grasping time window, and perform a Cartesian product operation on the projection area of ​​the three-dimensional motion trajectory of each end effector on the chessboard plane and the preset grasping time window to obtain the spatiotemporal occupancy area of ​​the end effector. The preset grasping time window is the time interval for all end effectors in the same collaborative grasping group to complete one grasping operation. S242: Calculate the degree of overlap of the spatiotemporal occupied regions of all end effectors corresponding to the two groups to obtain the overlap parameter; calculate the ratio of the overlap parameter to the total volume of all spatiotemporal occupied regions corresponding to the two groups to obtain the end effector interference probability between the two groups. S243: If the end effector interference probability exceeds the preset interference threshold, it is determined that there is interference between the two groups. The pieces in the two groups are first redistributed. If the end effector interference probability between the two groups still exceeds the preset interference threshold after redistribution, the two groups are merged.

[0011] Further, step S241, which involves performing a Cartesian product operation between the projection area of ​​the three-dimensional motion trajectory of each end effector on the chessboard plane and the preset grasping time window to obtain the spatiotemporal occupancy area of ​​the end effector, includes: S2411: For each end effector, sample along the three-dimensional motion trajectory of the end effector within a preset grasping time window at a preset sampling interval to obtain multiple trajectory sampling points. The preset sampling interval is determined based on the ratio of the chess piece diameter to the motion speed of the end effector. S2412: Project each trajectory sampling point vertically onto the chessboard plane to obtain a set of planar projection points. Construct a minimum bounding polygon for the set of planar projection points and use the minimum bounding polygon as the projection area of ​​the end effector on the chessboard plane. S2413: Perform a Cartesian product operation on the projected area and the preset capture time window to obtain the spatiotemporal occupancy area of ​​the end effector. The spatiotemporal occupancy area is a prismatic spatiotemporal region with the projected area as the base and the preset capture time window as the height.

[0012] Further, the preset envelope boundary distance threshold mentioned in step S222 is determined based on the end effector's end radius, including: S2221: Obtain the end shape parameters of the end effector and the cross-sectional shape parameters of the chess piece, and determine the shape compensation coefficient based on the end shape parameters of the end effector and the cross-sectional shape parameters of the chess piece. The shape compensation coefficient is used to characterize the degree of geometric adaptation when the end of the end effector contacts the edge of the chess piece. S2222: Obtain the number of pieces in the initial group, obtain the number of pieces per unit area of ​​the chessboard plane, wherein the number of pieces per unit area of ​​the chessboard plane is the ratio of the total number of remaining pieces on the chessboard to the chessboard plane area, obtain a preset reference area, wherein the preset reference area is a pre-set area benchmark value, and determine a density compensation coefficient based on the ratio of the number of pieces in the initial group to the number of pieces per unit area of ​​the chessboard plane and the ratio of the preset reference area, wherein the density compensation coefficient is used to characterize the additional influence of the density of pieces in the initial group on the safety margin of the spatial envelope boundary; S2223: Multiply the end radius of the end effector, the shape compensation coefficient, and the density compensation coefficient, and use the product as the preset envelope boundary distance threshold.

[0013] Further, step S3, which involves generating multi-end-effect collaborative grasping instructions for each of the collaborative grasping groups that match the number and spatial distribution of pieces within the collaborative grasping group, includes: S31: For each collaborative grasping group, obtain the static friction coefficient between each piece and the chessboard surface, the mass attribute of each piece, and the position of the piece assigned to be grasped by each end effector on the chessboard. S32: Calculate the micro-vibration sensitivity of each piece based on the static friction coefficient and the mass property. The micro-vibration sensitivity is used to characterize the sensitivity of the piece to positional displacement due to vibration transmission from the chessboard surface when other pieces in the same cooperative grasping group are contacted and grasped by the end effector. The micro-vibration sensitivity is negatively correlated with the static friction coefficient and is a dimensionless coefficient. S33: For any two pieces in the collaborative grasping group, calculate the vibration transmission coupling coefficient between the two pieces based on the minimum distance between them and the micro-vibration sensitivity, and construct a vibration transmission coupling matrix. Each element in the vibration transmission coupling matrix corresponds to the vibration transmission coupling coefficient between a pair of pieces, and the diagonal elements of the vibration transmission coupling matrix are zero. S34: The vibration transmission coupling matrix is ​​written as a feedforward compensation parameter into the multi-end effector collaborative grasping instruction. The feedforward compensation parameter is a parameter used to adjust the grasping force timing of the end effector before the grasping operation is executed. The multi-end effector collaborative grasping instruction also defines the grasping order, grasping path, grasping force parameter, and grasping force application time sequence of each end effector. The grasping force parameter includes the grasping force amplitude. The grasping force application time sequence is determined according to the vibration transmission coupling matrix, so that there is a preset time difference between the starting time of the grasping force application of each end effector in the same collaborative grasping group. The preset time difference is used to stagger the propagation of the chessboard surface vibration peak caused by the grasping force application. The preset grasping time window is the time interval for all end effectors in the same collaborative grasping group to complete one grasping operation.

[0014] Furthermore, the preset time difference mentioned in step S34 is determined according to the following steps: S3411: Obtain the vibration propagation speed of the chessboard surface material, wherein the vibration propagation speed is used to characterize the speed at which vibration energy propagates along the chessboard plane when the chessboard surface is subjected to the grasping force of the end effector. S3412: For any two end effectors in the same collaborative grasping group, obtain the minimum distance value between the chess pieces corresponding to the two end effectors respectively, and determine the end effector with the earlier grasping force application start time as the end effector that applies the grasping force first, and determine the end effector with the later grasping force application start time as the end effector that applies the grasping force later. S3413: Divide the minimum spacing value by the vibration propagation speed to obtain the vibration propagation time, which is used to characterize the time required for the vibration peak to propagate from one chess piece position to another chess piece position; S3414: Set the preset time difference to a value greater than the vibration propagation time, so that when the vibration peak generated by the end effector that applies the gripping force first propagates to the position of the chess piece corresponding to the end effector that applies the gripping force later, the end effector that applies the gripping force later has not yet started to apply the gripping force.

[0015] Furthermore, the time series of gripping force application described in step S34 is generated according to the following steps: S3421: Using the vibration transmission coupling matrix as input, calculate the vibration superposition amplitude at each chess piece position on the chessboard surface based on the vibration transmission coupling matrix and the grasping force amplitude in the grasping force parameters of all end effectors. The vibration superposition amplitude is the maximum value of the vibration displacement on the chessboard surface. S3422: Taking the minimum amplitude of the vibration superposition as the adjustment target, within the adjustable range of the starting time of the grasping force application of each end effector, the starting time of the grasping force application of each end effector is iteratively adjusted. The adjustable range is the time interval determined by the product of the micro-vibration sensitivity of the chess piece corresponding to the end effector and the preset grasping time window. The micro-vibration sensitivity is a dimensionless coefficient. S3423: After each iteration of adjustment, the vibration superposition amplitude is recalculated according to the vibration transmission coupling matrix, and the adjustment results that reduce the vibration superposition amplitude are retained until the vibration superposition amplitude no longer decreases. S3424: Determine the starting time of the gripping force application of each end effector after iterative adjustment as the gripping force application time sequence, and write it into the multi-end effector collaborative gripping instruction.

[0016] Accordingly, a second aspect of the present invention provides an electronic device, including: at least one processor; and a memory connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the above-described chess-playing interactive device chess-playing method based on multi-chess piece collaborative capture.

[0017] Accordingly, a third aspect of the present invention provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the above-described method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative capture.

[0018] The above-described technical solutions of the embodiments of the present invention have the following beneficial technical effects: 1. By constructing a piece state graph and introducing the condition that pieces in the same group can be grasped in parallel, the randomly scattered pieces on the chessboard are dynamically clustered and grouped, and the boundaries are adjusted. This ensures that the pieces in each collaborative grasping group satisfy the physical reach constraint of the maximum opening and closing distance of the end effector, and also ensures that the spatial envelope formed by the pieces in the group on the chessboard plane does not cover any unassigned pieces. This eliminates the potential mechanical interference caused by overlapping space occupancy when multiple end effectors grasp pieces simultaneously at the grouping level. This technique transforms the traditional piece-collecting method, which relies on fixed partitions or grasping pieces one by one, into a dynamic grouping mode based on the actual spatial distribution of pieces. This significantly improves the grouping rationality of parallel grasping by multiple end effectors and its adaptability to unstructured chessboard states. 2. By projecting the three-dimensional motion trajectory of the end effector within a preset grasping time window onto the chessboard plane and performing a Cartesian product with the time dimension, a spatiotemporal occupancy region is constructed. The ratio of the overlapping volume of the spatiotemporal occupancy regions corresponding to different groups to the total volume is used as a quantitative indicator of the end effector interference probability, transforming the interference risk from an empirical judgment into a calculable and comparable numerical basis. Based on this, an orderly adjustment operation of first redistribution and then merging is performed on groups with interference probabilities exceeding a preset threshold. After each operation, the interference probability of all groups is recalculated, forming a complete interference closed-loop control mechanism. This technique effectively solves the problems of unquantifiable interference risk and lack of deterministic rules for group adjustment in existing multi-end effect chess-collecting methods, significantly improving the conflict-free scheduling capability of parallel grasping of multiple pieces and the stability of the overall chess-collecting process. 3. When generating multi-end-effector collaborative grasping instructions, the micro-vibration sensitivity is calculated by obtaining the static friction coefficient and mass properties of each piece in the collaborative grasping group. A vibration transmission coupling matrix is ​​constructed based on the minimum spacing between pieces. Then, based on this matrix, a preset time difference between the starting times of the grasping force application of each end-effector in the same collaborative grasping group is determined to stagger the propagation of vibration peaks generated on the chessboard surface due to the application of grasping force. This technique is the first to incorporate the vibration transmission characteristics of the chessboard surface into the timing planning of multi-end-effector collaborative grasping. It actively suppresses the vibration superposition disturbance during the synchronous grasping of multiple pieces using a feedforward compensation method, avoiding piece position shifts and grasping failures caused by vibration transmission. This significantly improves the success rate of simultaneous grasping of multiple pieces and the adaptability to different chessboard materials and piece surface conditions. Attached Figure Description

[0019] Figure 1 This is a flowchart of the chess game collection method of the chess interactive device based on multi-chess piece collaborative capture provided in the embodiments of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0021] The chess-playing interaction device method based on multi-piece collaborative grasping described in this invention relies on a hardware architecture consisting of a vision unit, a control unit, and multiple end effectors. The vision unit, as the front-end sensing component for the chessboard state, acquires chessboard images and identifies the position, posture, type, and adjacency of the pieces, providing global state information for subsequent grouping and grasping. The control unit receives the piece state data provided by the vision unit, executes dynamic clustering grouping and collaborative grasping instruction generation, and uniformly schedules the actions of the multiple end effectors. The multiple end effectors respectively receive the grasping order, grasping path, and grasping force parameters issued by the control unit, and perform parallel grasping and transfer operations on multiple pieces within the same collaborative grasping group within the same time window. Compared with the single-endpoint actuator's piece-by-piece capture or fixed-partition scheduling method, this architecture enhances the multi-endpoint unified scheduling capability of the control unit and the global state perception capability of the vision unit in terms of functional division of labor. This enables multiple endpoint actuators to achieve on-demand grouping and parallel collaborative capture based on a unified description of the piece state graph, providing hardware support for dynamic grouping, interference constraints, and parallel execution during the piece collection process.

[0022] Please refer to Figure 1 The first aspect of this invention provides a method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative capture, comprising the following steps: S1: Use visual units to obtain the position, posture, type and adjacency relationship of all pieces on the chessboard, and construct a piece state graph based on position, posture, type and adjacency relationship. The adjacency relationship is the relationship that the minimum distance between two pieces does not exceed the preset adjacency distance threshold. The piece state graph is used to represent the spatial position relationship and state association of each piece on the chessboard and its adjacent pieces.

[0023] The vision unit acquires images of all pieces on the chessboard, obtaining the position, posture, and type of each piece, while extracting the adjacency relationship between any two pieces. Position refers to the specific landing point of the piece on the chessboard plane; posture reflects the angular deflection caused by placement or collision; and type distinguishes pieces of different colors, shapes, or materials. Adjacency is determined based on the minimum distance between two pieces. When this value does not exceed a preset adjacency distance threshold, the two pieces are considered to be adjacent. This threshold can be set according to the piece diameter and chessboard grid size to accommodate different chessboard sizes and piece combinations. The control unit organizes the position, posture, type, and adjacency relationship into a piece state map. This map records both the attributes of each piece and its spatial positional relationships and state associations with adjacent pieces. After the game, pieces are usually irregularly scattered, with some areas densely packed and others isolated. Some pieces deflect due to sliding or collision. The piece state map centrally expresses this scattered spatial and attribute information, eliminating the need for repeated image recognition and distance calculations in subsequent grouping.

[0024] S2: Based on the chess piece state graph, dynamically cluster and group the chess pieces on the chessboard to form multiple collaborative grasping groups. During dynamic clustering and grouping, the optimization objective is to minimize the grasping cost under the constraint that the interference probability of the end effector does not exceed a preset interference threshold. The end effector is the execution component in the chess interaction device used to grasp chess pieces; the grasping cost is the grasping cost value determined by the chess piece spacing and chess piece type; the interference probability is the probability that the movement spaces of multiple end effectors overlap; the preset interference threshold is the upper limit allowed for the probability of the end effector movement spaces overlapping; and a collaborative grasping group is a set of chess pieces consisting of two or more pieces that can be grasped in parallel by multiple end effectors within the same time window.

[0025] Unlike fixed partitioning, the grouping in this method is dynamically calculated based on the actual distribution of pieces on the current chessboard. The grouping process aims to minimize the capture cost while being constrained by the end-effector interference probability not exceeding a preset interference threshold. The capture cost is determined by the spacing and type of pieces; smaller spacing results in shorter end-effector movement paths and lower costs, while greater type differences increase the cost of adjusting capture force parameters or changing the capture configuration. The interference probability is the probability of overlapping motion spaces of multiple end-effectors, and the preset interference threshold is the upper limit of this probability. The grouping result must satisfy this constraint to ensure that the motion space does not overlap beyond the allowable level during parallel capture. A cooperative capture group consists of two or more pieces, and the pieces within the group can be captured in parallel by multiple end-effectors within the same time window. The combination of the above optimization objectives and constraints ensures that the grouping result considers both capture efficiency and execution feasibility, avoiding excessive overlap of motion spaces caused by simply pursuing compact grouping.

[0026] S3: For each collaborative grabbing group, generate multi-end-effector collaborative grabbing instructions that match the number and spatial distribution of pieces within the collaborative grabbing group. The multi-end-effector collaborative grabbing instructions limit the grabbing order, grabbing path, and grabbing force parameters of each end-effector for the corresponding piece.

[0027] For each collaborative grasping group, the control unit generates multi-end-effector collaborative grasping instructions that match the number and spatial distribution of pieces within the group. For each end-effector, the grasping order, grasping path, and grasping force parameters are specified. The grasping order coordinates the sequential actions of multiple end-effectors within the same group, ensuring that each end-effector approaches and grasps its corresponding piece in an orderly manner. The grasping path defines the movement route of the end-effector from its current position to above the target piece and complete the grasping action, avoiding the movement space of other end-effectors in the same group, ensuring that the movement trajectories of multiple end-effectors do not intersect within the same time window. The grasping force parameters are determined based on the type and material of the piece, avoiding excessive force that damages the piece or insufficient force that causes slippage. The number of pieces in the group determines the number of end-effectors participating in the grasping task, and the spatial distribution of the pieces determines the target position and movement direction of each end-effector. For cases where the types of pieces within the group are different, the grasping force parameters are set separately according to the material and surface characteristics of each piece.

[0028] S4: According to the multi-end-efficiency actuator collaborative grabbing instruction, control multiple end-efficiency actuators to simultaneously perform grabbing operations on pieces in the same collaborative grabbing group, and transfer the grabbed pieces to the collection area until all pieces on the board have been transferred.

[0029] Multiple end effectors simultaneously perform grasping operations on pieces within the same collaborative grasping group according to a multi-end effector collaborative grasping command. Each end effector arrives above its target piece according to the grasping order specified in the command, moves along the designated grasping path to the contact position, and clamps or attracts the piece with the corresponding grasping force parameters, transferring the grasped piece to the collection area. Multiple grasping actions within the same collaborative grasping group are executed in parallel within the same time window, allowing multiple pieces to be transferred at once, reducing the number of round trips between the end effectors and the collection area. After completing the transfer of the current collaborative grasping group, the control unit continues to perform grouping and grasping based on the updated state of the remaining pieces on the board until all pieces on the board have been transferred. The collection area can be a storage slot located on the side of the board. Multiple end effectors simultaneously transfer the pieces within the same collaborative grasping group to the storage slot and then release them, causing multiple pieces to fall into the storage slot at the same time.

[0030] By constructing a chess piece state graph, the randomly distributed chess pieces on the chessboard are dynamically clustered and grouped under the optimization conditions of minimizing the capture cost and the interference probability constraint. For each cooperative capture group, a multi-end effector cooperative capture instruction with limited capture order, capture path and capture force parameters is generated. Multiple end effectors can perform capture and transfer operations on multiple chess pieces in the same cooperative capture group in parallel within the same time window. The grouping rationality and parallel execution efficiency under the condition of unstructured chess piece distribution are improved. The problems of low efficiency of capture piece by piece, poor adaptability of fixed partitions and difficulty in effectively controlling the motion interference risk when multiple end effectors work in parallel are thus solved.

[0031] The specific implementation of dynamic clustering grouping in S2 is as follows: S21: Take all pieces in the piece state graph as objects to be grouped. Based on the adjacency relationship and spacing between pieces in the piece state graph, aggregate pieces that meet the preset proximity conditions in space to form multiple initial groups. The preset proximity condition is that the minimum spacing between two pieces does not exceed the maximum opening and closing distance of the end effector.

[0032] The piece state graph already stores whether any two pieces are adjacent and their corresponding minimum distance values. The control unit aggregates all pieces that meet the preset proximity condition to form multiple initial groups. The preset proximity condition is set so that the minimum distance between two pieces does not exceed the maximum opening and closing distance of the end effector. This ensures that any two pieces within the same initial group are physically within the range that a single end effector's opening and closing action can cover. For example, if there are five pieces close to each other in the upper left corner of the board, and the minimum distance between three of these pieces is less than the maximum opening and closing distance of the end effector, these three pieces are aggregated into one initial group. The remaining two pieces cannot be merged because their distance from these three pieces exceeds the maximum opening and closing distance of the end effector. They need to be aggregated separately with their respective neighboring pieces or remain as isolated pieces awaiting further processing. Isolated pieces whose distance from all other pieces exceeds this distance cannot form an initial group with any other piece. In subsequent processing, they are either captured individually by a single end effector or merged with other isolated pieces.

[0033] S22: For each initial group, determine whether the initial group satisfies the condition that the same group can be captured in parallel. The condition that the same group can be captured in parallel is that the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, and the area enclosed by the smallest bounding polygon formed by all pieces in the initial group on the chessboard plane does not contain other pieces not assigned to the initial group.

[0034] The condition for parallel grasping within the same group includes two requirements that must be met simultaneously. The first requirement is that the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, ensuring that the spatial distance between each piece in the group and other pieces in the same group is within the opening and closing stroke range of a single end effector. The second requirement is that the area enclosed by the smallest circumscribed polygon formed by all pieces in the initial group on the chessboard plane does not contain any other pieces not assigned to that initial group. This smallest circumscribed polygon is the smallest convex polygon determined based on the position coordinates of all pieces in the group, used to characterize the overall spatial occupancy range of the group of pieces on the chessboard plane. If three pieces in the initial group are located at the three vertices of a triangle, and there is another piece not belonging to that initial group inside the triangle area, then the unassigned piece is within the overall occupancy range of the group of pieces. The movement space and grasping path of the end effector may be obstructed when approaching its target piece, and the initial group is therefore deemed to have failed the condition.

[0035] S23: Perform a boundary adjustment operation on the initial group that does not meet the condition that the same group can be grabbed in parallel. The boundary adjustment operation is to remove the pieces that cause the area enclosed by the smallest bounding polygon to contain other pieces from the initial group, or to merge the pieces that are spatially continuous with the initial group and can be grabbed by the same end effector into the initial group. The piece type is the set of piece types that can be processed by the same end effector.

[0036] Boundary adjustment operations have two directions. The first direction removes pieces from the initial group that cause the area enclosed by the minimum bounding polygon to contain other pieces. When the position of a piece in a group causes the minimum bounding polygon to expand outwards and cover unassigned pieces, removing that piece shrinks the minimum bounding polygon formed by the remaining pieces in the group, preventing unassigned pieces from being covered. The second direction merges unassigned pieces that are spatially contiguous with the initial group and can be reliably grasped by the same end effector into the initial group. When an unassigned piece is located at the edge of the overall spatial footprint of the initial group and the minimum distance between it and at least one piece in the initial group does not exceed the maximum opening / closing distance of the end effector, the minimum bounding polygon corresponding to the newly formed set of pieces in the merged group may no longer contain any other unassigned pieces. Reliable grasping means that the type of unassigned pieces belongs to the same set of piece types that the end effector can handle, ensuring that the merged pieces can still be reliably grasped by the end effector corresponding to that group.

[0037] S24: For all groups formed after the boundary adjustment operation, calculate the end effector interference probability between any two groups. For two groups whose end effector interference probability exceeds the preset interference threshold, first redistribute the pieces in the two groups. If the end effector interference probability between the two groups still exceeds the preset interference threshold after redistribution, then merge the two groups. After each operation, recalculate the end effector interference probability between all pairs of groups until the end effector interference probability between all pairs of groups does not exceed the preset interference threshold. Output the final group as multiple collaborative grabbing groups.

[0038] When the end-effector interference probability between two groups exceeds a preset interference threshold, the control unit first redistributes the pieces in these two groups. For example, it redistributes a piece located at the boundary between the two groups from the first group to the second group, thus offsetting the projection areas of the end-effector movement spaces on the chessboard plane. If the interference probability still exceeds the preset threshold after redistribution, the two groups are merged, allowing end-effectors that might have caused spatial conflicts to be uniformly scheduled within the merged single group. After each redistribution or merging operation, the control unit recalculates the end-effector interference probability between all pairs of groups to check if the operation introduces new interference. This process is iterated until the interference probability between all pairs of groups does not exceed the preset interference threshold. The final output of multiple cooperative grasping groups is spatially independent and can perform grasping operations in parallel without interfering with each other during the same piece collection process.

[0039] The specific implementation of determining whether the initial group meets the condition that the same group can be crawled in parallel in S22 is as follows: S221: For each initial group, obtain the position coordinates of all pieces in the initial group on the chessboard plane, and determine the minimum bounding polygon of all pieces in the initial group. The area enclosed by the minimum bounding polygon is called the spatial envelope.

[0040] The minimum bounding polygon is the convex polygon with the smallest area that encloses the coordinates of all pieces in the group, and its vertices correspond to the actual positions of the pieces in the group. The spatial envelope does not directly refer to the actual shape of the pieces or the area of ​​the chessboard, but rather a geometric generalization of the overall area occupied by the pieces in the group on the chessboard plane. When there are three pieces in the group, and the three pieces are not on the same straight line, the spatial envelope is the triangular region with the positions of the three pieces as vertices; when there are only two pieces in the group, the spatial envelope simplifies to the line segment with the two pieces as endpoints and the narrow band region formed by the threshold expansion around it.

[0041] S222: For each other piece outside the initial group that is not assigned to the initial group, determine whether the position coordinates of the unassigned piece are located inside the spatial envelope, or whether the minimum distance between the position coordinates of the unassigned piece and the boundary of the spatial envelope is less than a preset envelope boundary distance threshold. The preset envelope boundary distance threshold is determined based on the end effector radius.

[0042] The judgment process comprises two levels: the first level determines whether the position coordinates of the unassigned piece are located within the spatial envelope; the second level determines whether the minimum distance between the unassigned piece's position coordinates and the boundary of the spatial envelope is less than a preset envelope boundary distance threshold. This preset boundary distance threshold is determined based on the end effector's radius and is used to reserve a certain safety margin outside the geometric boundary of the spatial envelope. Because the end effector itself has a certain radial dimension when grasping a piece, even if the unassigned piece is outside the spatial envelope but its distance from the boundary is less than the safety range corresponding to the end effector's radius, the unassigned piece may still interfere with the end effector's movement or be accidentally touched during parallel grasping due to excessive proximity. By setting this threshold, boundary proximity is also included in the judgment range for non-compliance, extending the spatial exclusivity constraint from inside the spatial envelope to a certain safety range outside the spatial envelope boundary.

[0043] S223: If the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, and the spatial envelope does not contain any other pieces not assigned to the initial group, and the minimum distance between the boundary of the spatial envelope and any unassigned piece is not less than the preset envelope boundary distance threshold, then the initial group is determined to meet the condition that the same group can be grasped in parallel; otherwise, the initial group is determined not to meet the condition that the same group can be grasped in parallel.

[0044] The judgment criteria include three requirements that must be met simultaneously: the minimum distance between any two pieces within a group does not exceed the maximum opening and closing distance of the end effector; the spatial envelope does not contain any unassigned pieces; and the minimum distance between the boundary of the spatial envelope and any unassigned piece is not less than a preset envelope boundary distance threshold. If any one of the three requirements is not met, the initial grouping will enter the boundary adjustment operation. Physical accessibility, spatial exclusivity, and boundary safety margin are incorporated into a unified judgment criterion, providing clear screening criteria for subsequent boundary adjustment and inter-group interference handling.

[0045] The specific implementation method of the boundary adjustment operation in S23 is as follows: S231: For an initial group that does not meet the condition that the same group can be grabbed in parallel, determine that the area enclosed by the smallest bounding polygon contains unassigned pieces within the group, and remove the pieces within the group from the initial group.

[0046] When an initial group fails to pass the judgment due to the presence of unassigned pieces within its spatial envelope, the control unit identifies a key piece among the pieces in the group that causes the region enclosed by the minimum bounding polygon to contain unassigned pieces. This key piece is typically characterized by its position on the chessboard plane significantly deviating from other pieces in the group, causing the minimum bounding polygon to expand outwards and cover unassigned pieces located between other pieces in the group. After removing this key piece, the positional distribution of the remaining pieces in the group is recalculated using the minimum bounding polygon, and the region enclosed by this polygon shrinks accordingly. The previously covered unassigned pieces no longer fall within the spatial envelope. The removed piece re-enters the grouping process as an unassigned piece and can subsequently be re-aggregated with other unassigned pieces or with adjacent groups.

[0047] S232: For unassigned pieces that are spatially contiguous with the initial group and can be captured by the same end effector, calculate the minimum distance between the unassigned pieces and the remaining pieces in the initial group. If the minimum distance does not exceed the maximum opening and closing distance of the end effector, then the unassigned pieces are merged into the initial group.

[0048] Spatial continuity means that the minimum distance between an unassigned piece and at least one piece in the initial group is within the range covered by the maximum opening and closing distance of the end effector. Compatible grasping means that the type of the unassigned piece belongs to the same set of piece types that the end effector can handle. For unassigned pieces that meet the conditions of spatial continuity and compatible grasping, the control unit calculates the minimum distance between the unassigned piece and the remaining pieces in the initial group. If this value does not exceed the maximum opening and closing distance of the end effector, the unassigned piece is merged into the initial group. After merging, the overall distribution of pieces in the group changes, and the corresponding minimum bounding polygon is recalculated. The new spatial envelope may no longer contain other unassigned pieces, thus ensuring that the initial group after merging meets the conditions.

[0049] S233: Repeat S231 and S232 until the initial group meets the condition that the same group can be crawled in parallel.

[0050] Since a single removal or merging operation may not necessarily ensure that the initial group meets the conditions, the control unit determines whether the initial group has met the condition for parallel capture within the same group based on the recalculated spacing between pieces, spatial envelope range, and boundary safety distance after each operation. If it still does not meet the condition, the removal or merging operation continues. The removal operation shrinks the spatial envelope, mainly used to eliminate conflicts involving unassigned pieces within the spatial envelope; the merging operation expands the spatial envelope, mainly used to incorporate unassigned pieces that are too close together outside the boundary into the group for unified scheduling. The two operations are executed alternately or sequentially in the iteration, gradually adjusting the spatial occupancy range of the initial group to a state where it does not interfere with surrounding unassigned pieces. If the condition still cannot be met after a preset number of adjustments, the initial group is split or merged with an adjacent group.

[0051] The specific implementation method for calculating the interference probability in S24 is as follows: S241: For any two groups, obtain the three-dimensional motion trajectory of the end effector corresponding to all pieces in the two groups within the preset grasping time window. Perform a Cartesian product operation on the projection area of ​​the three-dimensional motion trajectory of each end effector on the chessboard plane and the preset grasping time window to obtain the spatiotemporal occupancy area of ​​the end effector. The preset grasping time window is the time interval for all end effectors in the same collaborative grasping group to complete one grasping operation.

[0052] The three-dimensional motion trajectory is the complete spatial path of the end effector from its starting position to above the target piece and complete the grasping action. The control unit performs a Cartesian product operation on the projection area of ​​each end effector's three-dimensional motion trajectory on the chessboard plane and a preset grasping time window to obtain a three-dimensional spatial region with the chessboard plane projection area as the base and the preset grasping time window as the height direction, i.e., the spatiotemporal occupied region. This region describes the complete range of space occupied by the end effector above the chessboard within a specific time period. When the spatiotemporal occupied regions of two end effectors overlap, it indicates that the two end effectors may reach the same or similar positions on the chessboard plane within the same time period, and there is a potential possibility of motion interference.

[0053] S242: Calculate the degree of overlap of the spatiotemporal occupancy regions of all end effectors corresponding to the two groups to obtain the overlap parameter. Calculate the ratio of the overlap parameter to the total volume of all spatiotemporal occupancy regions corresponding to the two groups to obtain the end effector interference probability between the two groups.

[0054] For all end effectors in the two groups, the control unit pairs end effectors belonging to different groups and performs an overlap analysis on the spatiotemporal occupancy areas of each pair to determine the area they jointly cover in both spatial and temporal dimensions. The overlap ranges generated between each pair of end effectors are then summarized to obtain an overlap parameter. A larger overlap parameter indicates a higher degree of mutual intrusion in the motion space of the two groups' end effectors within the capture time window. The control unit simultaneously calculates the total volume of all spatiotemporal occupancy areas corresponding to the two groups and calculates the ratio of the overlap parameter to the total volume. The resulting ratio is the end effector interference probability between the two groups. This interference probability is directly derived from the spatiotemporal occupancy range of the actual motion trajectory of the end effectors, providing a quantifiable and comparable basis for the interference risk between groups.

[0055] S243: If the end effector interference probability exceeds the preset interference threshold, it is determined that there is interference between the two groups. The pieces in the two groups are redistributed. If the end effector interference probability between the two groups still exceeds the preset interference threshold after redistribution, the two groups are merged.

[0056] If the interference probability does not exceed a preset threshold, the two groups remain independent and execute their respective parallel grasping tasks. If the interference probability exceeds the preset threshold, the control unit first transfers key pieces located at the spatial boundary of the two groups or causing overlap in spatiotemporal occupancy areas from one group to the other, changing the distribution of the spatiotemporal occupancy areas of the end effectors corresponding to the two groups, thus reducing the degree of overlap. If the interference probability still exceeds the preset threshold after reallocation, the two groups are merged, so that the end effectors originally belonging to two groups are uniformly scheduled by the same collaborative grasping command within the merged single group. After each reallocation or merging operation, the control unit recalculates the interference probability between the affected group and other groups to avoid the operation causing new interference.

[0057] The specific implementation method for constructing the spatiotemporal occupancy region in S241 is as follows: S2411: For each end effector, sample along the three-dimensional motion trajectory of the end effector within a preset grasping time window at a preset sampling interval to obtain multiple trajectory sampling points. The preset sampling interval is determined based on the ratio of the chess piece diameter to the end effector's motion speed.

[0058] The preset sampling interval is determined based on the ratio of the chess piece diameter to the end effector's speed. If the sampling interval is too large, the trajectory sampling points are too sparse, resulting in a significant deviation between the subsequent projected area and the actual coverage area of ​​the end effector. If the sampling interval is too small, the number of trajectory sampling points increases dramatically, leading to a corresponding increase in computational load. By correlating the sampling interval with the ratio of the chess piece diameter to the end effector's speed, the control unit controls the overall number of trajectory sampling points while ensuring the trajectory discretization accuracy meets the requirements for interference judgment.

[0059] S2412: Project each trajectory sampling point vertically onto the chessboard plane to obtain a set of planar projection points. Construct a minimum bounding polygon for the set of planar projection points and use the minimum bounding polygon as the projection area of ​​the end effector on the chessboard plane.

[0060] The vertical projection operation maps each sampling point onto the chessboard plane along a direction perpendicular to the chessboard plane, and after removing the height component, obtains the planar projection points. After all trajectory sampling points are projected, a planar projection point set is formed, which centrally reflects the horizontal spatial coverage of the end effector when it moves above the chessboard plane within the preset grasping time window. The control unit constructs a minimum bounding polygon for this planar projection point set. This minimum bounding polygon is a convex polygon with the smallest area that includes all points in the planar projection point set, and its vertices are selected from the planar projection point set, with no indentations between the edges.

[0061] S2413: Perform a Cartesian product operation on the projected area and the preset capture time window to obtain the spatiotemporal occupancy area of ​​the end effector. The spatiotemporal occupancy area is a prism-shaped spatiotemporal region with the projected area as the base and the preset capture time window as the height.

[0062] Since the projection region is the smallest bounding polygon constructed by projecting all trajectory sampling points, the actual horizontal movement range of the end effector within the preset grasping time window will not exceed the boundary defined by the projection region. Therefore, the prismatic spatiotemporal region with the projection region as its base and the preset grasping time window as its height can completely cover the spatial occupancy range corresponding to the motion trajectory of the end effector within the time window. After the spatiotemporal occupancy regions of the end effectors in the two groups are constructed, the degree of motion interference between the groups can be quantified by performing overlap analysis on each prismatic spatiotemporal region.

[0063] The method for determining the preset envelope boundary distance threshold in S222 is as follows: S2221: Obtain the end shape parameters of the end effector and the cross-sectional shape parameters of the chess piece. Determine the shape compensation coefficient based on the end shape parameters of the end effector and the cross-sectional shape parameters of the chess piece. The shape compensation coefficient is used to characterize the degree of geometric fit when the end of the end effector contacts the edge of the chess piece.

[0064] When the shape of the contact surface at the end of the end effector matches the cross-sectional shape of the chess piece well, the effective contact range between the end effector and the edge of the chess piece during the grasping process is large, and the actual safety margin requirement between the outer edge of the end effector and the spatial envelope boundary is relatively small, so the shape compensation coefficient takes a small value. When the shape of the end effector and the cross-sectional shape of the chess piece are poorly matched, its end edge may extend a larger range outside the chess piece due to shape mismatch. In this case, a larger envelope boundary safety distance is required to avoid unexpected contact between the edge of the end effector and the unassigned chess piece, so the shape compensation coefficient takes a larger value accordingly.

[0065] S2222: Obtain the number of pieces in the initial group, obtain the number of pieces per unit area of ​​the chessboard plane, the number of pieces per unit area of ​​the chessboard plane is the ratio of the total number of remaining pieces on the chessboard to the chessboard plane area, obtain the preset reference area, the preset reference area is a pre-set area benchmark value, determine the density compensation coefficient based on the ratio of the number of pieces in the initial group to the number of pieces per unit area of ​​the chessboard plane and the preset reference area, the density compensation coefficient is used to characterize the additional influence of the density of pieces in the initial group on the safety margin of the spatial envelope boundary.

[0066] The number of pieces per unit area on the chessboard reflects the distribution density of the remaining pieces in the overall space. The ratio of the number of pieces in the initial group to the number of pieces per unit area on the chessboard has area dimensions. This ratio is eliminated by the ratio of the preset reference area, making the density compensation coefficient dimensionless. When the remaining pieces on the chessboard are sparsely distributed, the number of pieces per unit area is small, the area ratio corresponding to the number of pieces in the initial group is relatively large, the density compensation coefficient takes a larger value, and the envelope boundary distance threshold increases accordingly. When the remaining pieces on the chessboard are densely distributed, the number of pieces per unit area is large, the area ratio corresponding to the number of pieces in the initial group is relatively small, the density compensation coefficient takes a smaller value, avoiding excessive expansion of the spatial envelope range in dense areas due to excessively large safety distance settings.

[0067] S2223: Multiply the end radius of the end effector, the shape compensation coefficient, and the density compensation coefficient, and use the product as the preset envelope boundary distance threshold.

[0068] The end effector's end radius has a length dimension, while the shape compensation coefficient and density compensation coefficient are dimensionless coefficients. The product of these three factors has a length dimension. The end effector's end radius, as a fundamental quantity, directly reflects the radial range occupied by the outer edge of the end effector relative to its central axis when the end effector moves above the chessboard plane. The shape compensation coefficient amplifies or reduces the safety distance based on the geometric fit between the end effector and the chess piece. The density compensation coefficient further adjusts the safety distance based on the overall chess piece density of the chessboard. The product of these three factors comprehensively reflects the combined influence of the end effector's own dimensions, the shape matching relationship between the end effector and the chess piece, and the current chess piece distribution density on the spatial envelope boundary safety distance.

[0069] The specific implementation method of multi-end-effect actuator collaborative fetching instruction generation in S3 is as follows: S31: For each collaborative grabbing group, obtain the static friction coefficient between each piece and the chessboard surface, the mass attribute of each piece, and the position of the piece assigned to be grabbed by each end effector on the chessboard.

[0070] The static friction coefficient reflects the ability of a chess piece's bottom surface to resist relative sliding when in contact with the chessboard surface. It is directly related to the material of the chess piece, the material of the chessboard surface, and the contact state between them. The mass property reflects the magnitude of the chess piece's own inertia. In practical devices, the static friction coefficient can be obtained by pre-calibrating the combination of chess piece and chessboard materials, and the static friction coefficients corresponding to different material combinations are stored in the parameter table of the control unit. During the game, the coefficient is directly called based on the actual materials of the current chessboard and chess pieces.

[0071] S32: Calculate the micro-vibration sensitivity of each piece based on the static friction coefficient and mass properties. The micro-vibration sensitivity is used to characterize the degree to which a piece is sensitive to positional displacement due to vibration transmission from the chessboard surface when other pieces in the same cooperative grasping group are contacted by the end effector and grasping force is applied. The micro-vibration sensitivity is negatively correlated with the static friction coefficient and is a dimensionless coefficient.

[0072] When the end effector applies a grasping force to a piece, the force generated at the moment of contact is transmitted through the piece to the board surface, creating local vibrations. These vibrations propagate along the board plane and may cause slight positional shifts in other pieces within the same group. A higher static friction coefficient results in stronger static resistance between the piece and the board surface, making it less prone to positional shifts under vibration disturbances and exhibiting lower micro-vibration sensitivity. Conversely, a lower static friction coefficient makes the piece more susceptible to sliding or shifting under vibration disturbances, leading to higher micro-vibration sensitivity. The influence of mass on micro-vibration sensitivity manifests as an inertial effect; pieces with greater mass, under the same vibration excitation, experience relatively smaller acceleration responses and positional shifts due to their greater inertia. Micro-vibration sensitivity is only used for relative comparison and ranking of the vibration response characteristics of different pieces within the same cooperative grasping group, providing a normalized basis for constructing the subsequent vibration transmission coupling matrix.

[0073] S33: For any two pieces in the collaborative capture group, calculate the vibration transmission coupling coefficient between any two pieces based on the minimum distance between them and the micro-vibration sensitivity, and construct a vibration transmission coupling matrix. Each element in the vibration transmission coupling matrix corresponds to the vibration transmission coupling coefficient between a pair of pieces, and the diagonal elements of the vibration transmission coupling matrix are zero.

[0074] The vibration transmission coupling coefficient characterizes the degree of correlation between two pieces when the end effector of one piece applies a grasping force, resulting in a disturbance in the other piece caused by the vibration transmitted through the chessboard surface. The smaller the minimum distance between the two pieces, the shorter the distance the vibration travels from one piece's position to the other, the less the vibration energy attenuates during propagation, and the higher the vibration transmission coupling coefficient. The higher the micro-vibration sensitivity of the disturbed piece, the more pronounced its positional shift response under the same vibration input, and the higher the vibration transmission coupling coefficient between them. The diagonal elements correspond to the coupling relationship between a piece and itself. Since the piece's own grasping action does not constitute a disturbance transmitted through the chessboard surface, the diagonal elements are zero.

[0075] S34: The vibration transmission coupling matrix is ​​written as a feedforward compensation parameter into the multi-end effector collaborative grasping instruction. The feedforward compensation parameter is used to adjust the grasping force timing of the end effector before the grasping operation is executed. The multi-end effector collaborative grasping instruction also defines the grasping order, grasping path, grasping force parameter and grasping force application time sequence of each end effector. The grasping force parameter includes the grasping force amplitude. The grasping force application time sequence is determined according to the vibration transmission coupling matrix, so that there is a preset time difference between the starting time of the grasping force application of each end effector in the same collaborative grasping group. The preset time difference is used to stagger the propagation of the chessboard surface vibration peak caused by the grasping force application. The preset grasping time window is the time interval for all end effectors in the same collaborative grasping group to complete one grasping operation.

[0076] The purpose of feedforward compensation parameters is to pre-adjust the application time sequence of each end effector's grasping force based on the vibration transmission relationship between pieces within the group before executing the grasping action. When multiple end effectors in the same cooperative grasping group apply grasping force to their respective target pieces simultaneously or almost simultaneously, the vibrations generated by each grasping action will propagate along the chessboard surface. If multiple vibration peaks arrive at the same piece's position at the same time, the disturbance experienced by that piece will be significantly amplified. By determining the preset time difference between the start times of the grasping force application of each end effector based on the vibration transmission coupling matrix, the vibration peaks generated by each grasping action are staggered during propagation, reducing the superposition of vibration energy arriving at the same position at the same time.

[0077] The preset time difference is determined in S34 as follows: S3411: Obtain the vibration propagation speed of the chessboard surface material. The vibration propagation speed is used to characterize how fast the vibration energy of the chessboard surface propagates along the chessboard plane when it is subjected to the grasping force of the end effector.

[0078] The specific value of vibration propagation velocity depends on factors such as the elastic modulus, density, and thickness of the chessboard material. In practical applications, it can be obtained by pre-calibrating the vibration characteristics of the chessboard material and storing the calibration results in the control unit. The purpose of obtaining the vibration propagation velocity is to provide a physical basis for the subsequent calculation of vibration propagation time, so that the determination of the preset time difference can be quantified based on the actual vibration characteristics of the chessboard material, rather than using a fixed time value unrelated to the chessboard material.

[0079] S3412: For any two end effectors in the same collaborative grasping group, obtain the minimum distance value between the pieces corresponding to the two end effectors respectively, and determine the end effector with the earlier grasping force application start time as the end effector that applies the grasping force first, and determine the end effector with the later grasping force application start time as the end effector that applies the grasping force later.

[0080] The determination of this sequence is used to clarify the direction and starting point of vibration propagation. That is, the position of the chess piece corresponding to the end effector that applies the gripping force first is the starting position of vibration excitation, and the position of the chess piece corresponding to the end effector that applies the gripping force later is the target position of vibration propagation. When the starting time of the gripping force application of the two end effectors has not yet been finalized, the control unit can make a sequence judgment based on the arrangement in the initial time sequence, and reorder them according to the actual determined time sequence during the subsequent adjustment process.

[0081] S3413: Divide the minimum spacing value by the vibration propagation speed to obtain the vibration propagation time. The vibration propagation time is used to characterize the time required for the vibration peak to propagate from one piece position to another.

[0082] When the end effector that applies the gripping force first applies the gripping force to the corresponding piece, a vibration peak is generated at the piece's location. The vibration peak propagates outwards along the chessboard surface, reaching the position of the piece corresponding to the end effector that applies the gripping force later after a certain propagation time. The magnitude of the vibration propagation time depends on the distance between the two pieces and the vibration propagation speed of the chessboard material; the greater the distance, the longer the propagation time; the faster the vibration propagation speed, the shorter the propagation time.

[0083] S3414: Set the preset time difference to a value greater than the vibration propagation time, so that when the vibration peak generated by the end effector that applies the gripping force first propagates to the position of the piece corresponding to the end effector that applies the gripping force later, the end effector that applies the gripping force later has not yet started to apply the gripping force.

[0084] When the preset time difference is greater than the vibration propagation time, the vibration peak generated by the end effector that applies the gripping force first will reach the corresponding piece position of the end effector that applies the gripping force later, before the latter end effector has even begun applying its gripping force. Because the preset time difference is strictly greater than the vibration propagation time, the moment the latter end effector begins applying its gripping force is later than the moment the vibration peak generated by the first application reaches its corresponding piece position, thus avoiding temporal overlap between the vibration peaks generated by the first and latter applications at the same piece position. For cases where the cooperative gripping group contains more than three pieces, the control unit determines the preset time difference for each pair of end effectors with a sequential gripping relationship using this method.

[0085] The time series of gripping force application in S34 is generated as follows: S3421: Using the vibration transmission coupling matrix as input, the vibration superposition amplitude at each chess piece position on the chessboard surface is calculated based on the vibration transmission coupling matrix and the grasping force amplitude in the grasping force parameters of all end effectors. The vibration superposition amplitude is the maximum value of the vibration displacement on the chessboard surface.

[0086] The larger the grasping force amplitude, the stronger the excitation energy transmitted to the chessboard surface when the end effector contacts the piece, and the higher the vibration response amplitude generated on the chessboard surface. The control unit distributes the grasping force amplitude of each end effector to the positions of other pieces in the same group according to the coupling coefficient in the vibration transmission coupling matrix, obtaining the vibration contribution components from the grasping actions of different end effectors in the same group at each piece position. All vibration contribution components at the same piece position are then superimposed to obtain the superimposed vibration amplitude at that piece position.

[0087] S3422: With the minimum amplitude of vibration superposition as the adjustment target, within the adjustable range of the starting time of the grasping force application of each end effector, the starting time of the grasping force application of each end effector is iteratively adjusted. The adjustable range is the time interval determined by the product of the micro-vibration sensitivity of the chess piece corresponding to the end effector and the preset grasping time window. The micro-vibration sensitivity is a dimensionless coefficient.

[0088] When the starting times of the gripping forces applied by multiple end effectors are the same or close, the vibration peaks generated by each gripping action are prone to arrive at the same piece position at the same time during propagation and thus superimpose. When the starting times of the gripping forces are staggered, the arrival times of the vibration peaks at the same piece position are dispersed, and the degree of superposition is reduced. The higher the micro-vibration sensitivity, the more easily the piece is deflected by vibration, and the larger the adjustable range of the starting time of the gripping force applied by the corresponding end effector is, so as to have a larger adjustment space on the time axis to avoid vibration peaks.

[0089] S3423: After each iteration adjustment, recalculate the vibration superposition amplitude based on the vibration transmission coupling matrix, retain the adjustment results that reduce the vibration superposition amplitude, until the vibration superposition amplitude no longer decreases.

[0090] If the adjusted vibration superposition amplitude is smaller than the original amplitude, it indicates that the adjustment reduced the degree of vibration superposition, and the control unit retains the adjustment result. If the adjusted vibration superposition amplitude does not decrease, the adjustment is abandoned, and the timing state is restored to its original state. This iterative process terminates when the vibration superposition amplitude no longer decreases. When vibration superposition exists simultaneously at multiple piece positions, the control unit can use the maximum value of the vibration superposition amplitude at all piece positions as the evaluation index for iterative adjustment.

[0091] S3424: Determine the starting time of the gripping force application of each end effector after iterative adjustment as the gripping force application time sequence, and write it into the multi-end effector collaborative gripping instruction.

[0092] Through iterative optimization, the distribution of the starting time of the grasping force application of each end effector on the time axis has minimized the superposition amplitude of vibrations at the positions of each piece on the chessboard surface, and the grasping actions of each end effector within the group have formed a staggered temporal relationship. This grasping force application time sequence, together with the grasping order, grasping path, and grasping force parameters of each end effector, constitutes the complete content of the multi-end effector collaborative grasping instruction.

[0093] The following are three specific embodiments.

[0094] Example 1 This embodiment applies to a Go teaching and training device. The device includes a vision unit, a control unit, and three end effectors. Each end effector is mounted on a horizontally movable slide mechanism above the Go board. Each end effector employs a two-finger gripper structure, has a 6 mm end radius, and a maximum opening / closing distance of 30 mm. The Go board is a wooden Go board with 22 mm grid dimensions. The Go pieces are double-sided convex ceramic pieces with a diameter of 21 mm. The vision unit is an industrial camera mounted 60 cm directly above the Go board.

[0095] Before the game begins, several Go stones are scattered on the board, randomly distributed. Some stones are adjacent to each other, some are isolated, and some have deflected due to sliding. The vision unit acquires a complete image of the board surface, extracting the position coordinates of each stone through edge detection and circular fitting algorithms. The stone's posture is determined by its surface reflectivity and contour direction, and black and white stones are distinguished by their color and surface texture. Simultaneously, the vision unit calculates the minimum distance between any two stones, converting the minimum pixel distance between the outlines of two stones in the image into the actual physical distance. The control unit sets a preset adjacency distance threshold of 3 mm based on the stone diameter of 21 mm and the board grid size of 22 mm. All identified stone positions, postures, types, and adjacency relationships are uniformly constructed into a stone state map, stored in the control unit's memory as a data table. Each stone corresponds to a record, including the stone number, position coordinates, posture angle, type identifier, quality attribute, and a list of adjacent stone numbers.

[0096] The control unit performs dynamic clustering and grouping based on the piece state map, aggregating pieces whose minimum distance between two pieces does not exceed 30 mm (the maximum opening and closing distance of the end effector) to form initial groups. In the upper left corner of the chessboard, there are 5 pieces. Three of these pieces have edge distances of 5 mm, 8 mm, and 12 mm between each pair, all not exceeding 30 mm, and are aggregating into an initial group. The other two pieces have minimum distances of 35 mm and 40 mm from these three pieces, exceeding 30 mm, and are either aggregated with nearby pieces or treated as isolated pieces awaiting further processing. Taking the initial group of these three pieces as an example, the control unit confirms that the minimum distance between any two of the three pieces does not exceed 30 mm, determining that the smallest bounding polygon formed by the three pieces is a triangular region. If there are no unassigned pieces within the triangular region, the initial group is retained as a cooperative grabbing group. If there is an unassigned piece within the triangular region, the control unit will remove pieces within the group that cover the unassigned piece, or merge unassigned pieces located at the edge of the triangular region that can be grabbed by the same end effector into the initial group, until the adjusted group meets the condition for parallel grabbing within the same group. After all initial groups have completed judgment and boundary adjustment, the control unit calculates the end effector interference probability between groups. For two groups whose interference probability exceeds a preset threshold, the pieces are first redistributed. If the probability still exceeds the threshold after redistribution, the two groups are merged, until the interference probability between all pairs of groups does not exceed the preset threshold.

[0097] For each collaborative grasping group, the control unit specifies the grasping order, grasping path, and grasping force parameters for each end effector. The three end effectors approach their respective target pieces sequentially from left to right, with grasping paths that do not intersect on the chessboard plane. The grasping force parameters are determined based on the contact characteristics between the ceramic pieces and the wooden chessboard surface. Within the same time window, each of the three end effectors simultaneously performs a grasping operation on three pieces. After the grippers close, they lift up, transferring all three pieces simultaneously to the receiving slots on the side of the chessboard. The grippers then release synchronously, causing all three pieces to fall into the receiving slots simultaneously. After completing the transfer of this group of pieces, the control unit continues grouping and grasping based on the updated state of the remaining pieces on the chessboard, until all pieces on the chessboard have been transferred to the receiving slots.

[0098] Example 2 This embodiment applies to a Chinese chess training device. The device includes a vision unit, a control unit, and four end effectors. Each end effector is mounted on a portal frame that can move horizontally above the chessboard. Each end effector uses a vacuum suction cup structure with a suction cup end radius of 4 mm and a maximum opening / closing distance of 20 mm. The chessboard is an engineering plastic chessboard with a grid size of 28 mm. The chess pieces are cylindrical plastic pieces with a diameter of 26 mm and a height of 10 mm. The static friction coefficient between the bottom surface of the chess piece and the chessboard surface is calibrated to 0.28. The mass of the chess pieces is weighed and recorded in the control unit. Before the chess piece collection operation begins, there are 20 pieces scattered on the chessboard, one for each side (red and black), with some areas having multiple pieces closely packed together.

[0099] Unlike Embodiment 1, this embodiment introduces a chessboard surface vibration transmission suppression mechanism when generating multi-end effector collaborative grasping instructions. The control unit obtains the static friction coefficient and mass properties of each chess piece in the group and calculates the micro-vibration sensitivity of each piece. Plastic chess pieces with a static friction coefficient of 0.28 are more prone to sliding when subjected to vibration disturbances on the chessboard surface, and their micro-vibration sensitivity is relatively high. The control unit calculates the vibration transmission coupling coefficient for any two chess pieces in the group and constructs a vibration transmission coupling matrix, with the diagonal elements of the matrix set to 0. Taking a collaborative grasping group containing 3 chess pieces as an example, the 3 suction cup ends correspond to 3 chess pieces respectively. The control unit uses the vibration transmission coupling matrix as a feedforward compensation parameter to determine that the first suction cup end applies the suction force first, the second suction cup end applies the suction force after a first preset time difference, and the third suction cup end applies the suction force after a second preset time difference. Each time difference is greater than the time obtained by dividing the distance between the corresponding chess pieces by the chessboard vibration propagation speed. The three suction cups adsorb the corresponding chess pieces at the same time window according to the instructions. After adsorption, they are simultaneously lifted and transferred to the chess piece collection area. Because the adsorption forces of each suction cup are applied at staggered times, the propagation of vibration peaks on the chessboard surface during the grasping process is effectively staggered, avoiding positional shifts of multiple chess pieces due to the superposition of vibrations.

[0100] During the grouping phase, six pieces are closely arranged in the right-hand area of ​​the chessboard. Four of these pieces have an edge spacing between each pair of pieces between 2 and 15 millimeters, and are grouped into an initial group. The control unit confirms that the minimum distance between any two pieces within the group does not exceed 20 millimeters. Based on the position coordinates of the four pieces, the minimum bounding polygon is determined to be an irregular quadrilateral. Within this quadrilateral area, there is one unassigned piece. The control unit removes the piece that is furthest from the other pieces in the group and causes the quadrilateral area to expand from the initial group. After removal, the minimum bounding polygon area formed by the remaining three pieces shrinks, no longer covering any unassigned pieces. The adjusted grouping satisfies the condition for parallel capture within the same group. After all 20 pieces have been transferred, the piece collection process ends.

[0101] Example 3 This embodiment applies to a multifunctional chess entertainment device compatible with various chess games such as Go, Gomoku, and Chinese Chess. The device is equipped with a vision unit, a control unit, and two end effectors. The two end effectors are mounted on a dual-arm robotic arm above the chessboard. Each end effector employs a flexible three-finger gripper structure, with an end radius of 8 mm and a maximum opening distance of 40 mm. The chessboard is made of composite material, with each grid measuring 25 mm. The device can use Go pieces with a diameter of 22 mm and Chinese Chess pieces with a diameter of 24 mm. The control unit stores a table of physical attribute parameters for various chess pieces. Before the game begins, different types of chess pieces are scattered on the board, and some pieces have shifted their posture due to collisions.

[0102] The focus of this embodiment is on the grouping compatibility handling of mixed-type chess pieces. There are four chess pieces in the central area of ​​the chessboard. Two Go pieces and one Chinese chess piece have an edge spacing of no more than 40 mm between each pair of them, and are grouped into an initial group. The remaining Go piece has an edge spacing of 45 mm with these three pieces, which does not meet the preset proximity condition and is processed separately. This initial group contains both Go and Chinese chess pieces. The control unit determines that the minimum distance between any two pieces in the group does not exceed 40 mm and determines the minimum bounding polygon based on the position coordinates of the three pieces. Since the diameter of the Go piece is different from that of the Chinese chess piece, the determination of the minimum bounding polygon is based on the outer edge of each piece's outline on the chessboard plane. The control unit checks whether there are any unassigned pieces within the minimum bounding polygon area. If not, it further determines whether the minimum distance between the spatial envelope boundary and any unassigned piece is not less than a preset envelope boundary distance threshold. This threshold is determined based on the end effector's end radius of 8 mm and is adjusted using shape compensation and density compensation coefficients. The shape compensation coefficient is determined based on the degree of matching between the end-effector's flexible three-finger gripper shape and the cross-sectional shape of the chess piece. A smaller value is used for cylindrical chess pieces with regular cross-sectional shapes, and a larger value is used for Go pieces with double-convex shapes. The density compensation coefficient is determined based on the ratio of the total number of remaining chess pieces on the board to the board's surface area and the number of chess pieces in the initial group.

[0103] During the gripping instruction generation phase, the flexible three-finger gripper applies different gripping forces to Go pieces and chess pieces respectively. Go pieces have a lower surface friction coefficient, so the gripping force is slightly larger to ensure stable gripping; chess pieces have a larger mass, so the gripping force is increased accordingly based on their mass. If a group contains three pieces and there are two available end effectors, the control unit assigns two of the three pieces to the two end effectors for parallel gripping. The third piece is gripped in the next time window, or it can be reassigned to another cooperative gripping group during the grouping adjustment phase. The control unit calculates the micro-vibration sensitivity based on the static friction coefficient and mass properties of the pieces in the group, constructs a vibration transmission coupling matrix, and determines that the preset time difference between the application of the gripping force between the two end effectors is greater than the vibration propagation time. This ensures that when the vibration peak generated by the gripping force applied by the first end effector propagates to the corresponding piece position of the second end effector, the second end effector has not yet begun applying its gripping force. Two end effectors apply gripping force in staggered shifts according to instructions, avoiding the impact of superimposed vibrations on the chessboard surface on the stability of the pieces' positions and improving the success rate of parallel gripping of mixed-type pieces.

[0104] Accordingly, a second aspect of the present invention provides an electronic device, including: at least one processor; and a memory connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the above-described chess-playing interactive device chess-playing method based on multi-chess piece collaborative capture.

[0105] Accordingly, a third aspect of the present invention provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the above-described method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative capture.

[0106] The embodiments of the present invention aim to protect a method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative capture, which has the following effects: 1. By constructing a piece state graph and introducing the condition that pieces in the same group can be grasped in parallel, the randomly scattered pieces on the chessboard are dynamically clustered and grouped, and the boundaries are adjusted. This ensures that the pieces in each collaborative grasping group satisfy the physical reach constraint of the maximum opening and closing distance of the end effector, and also ensures that the spatial envelope formed by the pieces in the group on the chessboard plane does not cover any unassigned pieces. This eliminates the potential mechanical interference caused by overlapping space occupancy when multiple end effectors grasp pieces simultaneously at the grouping level. This technique transforms the traditional piece-collecting method, which relies on fixed partitions or grasping pieces one by one, into a dynamic grouping mode based on the actual spatial distribution of pieces. This significantly improves the grouping rationality of parallel grasping by multiple end effectors and its adaptability to unstructured chessboard states. 2. By projecting the three-dimensional motion trajectory of the end effector within a preset grasping time window onto the chessboard plane and performing a Cartesian product with the time dimension, a spatiotemporal occupancy region is constructed. The ratio of the overlapping volume of the spatiotemporal occupancy regions corresponding to different groups to the total volume is used as a quantitative indicator of the end effector interference probability, transforming the interference risk from an empirical judgment into a calculable and comparable numerical basis. Based on this, an orderly adjustment operation of first redistribution and then merging is performed on groups with interference probabilities exceeding a preset threshold. After each operation, the interference probability of all groups is recalculated, forming a complete interference closed-loop control mechanism. This technique effectively solves the problems of unquantifiable interference risk and lack of deterministic rules for group adjustment in existing multi-end effect chess-collecting methods, significantly improving the conflict-free scheduling capability of parallel grasping of multiple pieces and the stability of the overall chess-collecting process. 3. When generating multi-end-effector collaborative grasping instructions, the micro-vibration sensitivity is calculated by obtaining the static friction coefficient and mass properties of each piece in the collaborative grasping group. A vibration transmission coupling matrix is ​​constructed based on the minimum spacing between pieces. Then, based on this matrix, a preset time difference between the starting times of the grasping force application of each end-effector in the same collaborative grasping group is determined to stagger the propagation of vibration peaks generated on the chessboard surface due to the application of grasping force. This technique is the first to incorporate the vibration transmission characteristics of the chessboard surface into the timing planning of multi-end-effector collaborative grasping. It actively suppresses the vibration superposition disturbance during the synchronous grasping of multiple pieces using a feedforward compensation method, avoiding piece position shifts and grasping failures caused by vibration transmission. This significantly improves the success rate of simultaneous grasping of multiple pieces and the adaptability to different chessboard materials and piece surface conditions.

[0107] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0109] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0110] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping, characterized in that, The chess-playing interactive device includes a vision unit, a control unit, and multiple end effectors. The vision unit is used to acquire chessboard images, the control unit is used to generate collaborative grasping instructions, and the multiple end effectors are used to grasp chess pieces. The method includes the following steps: S1: The visual unit is used to obtain the position, posture, type and adjacency relationship between all pieces on the chessboard, and a piece state graph is constructed based on the position, posture, type and adjacency relationship. The adjacency relationship is the relationship that the minimum distance between two pieces does not exceed a preset adjacency distance threshold. The piece state graph is used to characterize the spatial position relationship and state association between each piece on the chessboard and its adjacent pieces. S2: Based on the chess piece state graph, the chess pieces on the chessboard are dynamically clustered and grouped to form multiple collaborative grasping groups; when performing the dynamic clustering and grouping, the optimization objective is to minimize the grasping cost under the constraint that the interference probability of the end effector does not exceed a preset interference threshold. The end effector is the execution component in the chess interaction device used to grasp chess pieces. The grasping cost is the grasping cost value determined by the chess piece spacing and chess piece type. The interference probability is the probability that the movement spaces of multiple end effectors overlap. The preset interference threshold is the upper limit of the allowed probability that the movement spaces of end effectors overlap. The collaborative grasping group is a set of chess pieces consisting of two or more chess pieces that can be grasped in parallel by multiple end effectors in the same time window. S3: For each of the cooperative grasping groups, generate a multi-end-efficiency actuator cooperative grasping instruction that matches the number and spatial distribution of pieces within the cooperative grasping group. The multi-end-efficiency actuator cooperative grasping instruction defines the grasping order, grasping path, and grasping force parameters of each of the multiple end-efficiencies for the corresponding pieces. S4: According to the multi-end-efficiency actuator collaborative grasping instruction, control the multiple end-efficiency actuators to simultaneously perform grasping operations on the pieces in the same collaborative grasping group, and transfer the grasped pieces to the collection area until all pieces on the board have been transferred.

2. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 1, characterized in that, Step S2 describes dynamically clustering and grouping the pieces on the chessboard based on the piece state graph to form multiple collaborative fetching groups, including: S21: Take all the pieces in the chess piece state map as the objects to be grouped, and according to the adjacency relationship and spacing between the pieces in the chess piece state map, aggregate the pieces that meet the preset proximity conditions in space to form multiple initial groups. The preset proximity condition is that the minimum spacing between two pieces does not exceed the maximum opening and closing distance of the end effector. S22: For each initial group, determine whether the initial group satisfies the condition that the same group can be grabbed in parallel. The condition that the same group can be grabbed in parallel is that the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, and the area enclosed by the smallest bounding polygon formed by all the pieces in the initial group on the chessboard plane does not contain other pieces not assigned to the initial group. S23: Perform a boundary adjustment operation on the initial group that does not meet the condition that the same group can be grabbed in parallel. The boundary adjustment operation is to remove the pieces that cause the area enclosed by the minimum bounding polygon to contain other pieces from the initial group, or to merge the pieces that are spatially continuous with the initial group and can be grabbed by the same end effector into the initial group. The compatible grabbing means that the piece type belongs to the set of piece types that can be processed by the same end effector. S24: For all groups formed after the boundary adjustment operation, calculate the end effector interference probability between any two groups. For two groups whose end effector interference probability exceeds a preset interference threshold, first redistribute the pieces in the two groups. If the end effector interference probability between the two groups still exceeds the preset interference threshold after redistribution, then merge the two groups. After each operation, recalculate the end effector interference probability between all pairs of groups until the end effector interference probability between all pairs of groups does not exceed the preset interference threshold. Output the final group as multiple collaborative grabbing groups.

3. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 2, characterized in that, Step S22, determining whether the initial group satisfies the condition that the same group can be crawled in parallel, includes: S221: For each initial group, obtain the position coordinates of all pieces in the initial group on the chessboard plane, and determine the minimum bounding polygon of all pieces in the initial group. The area enclosed by the minimum bounding polygon is referred to as the spatial envelope. S222: For each other piece outside the initial group that is not assigned to the initial group, determine whether the position coordinates of the unassigned piece are located in the inner region of the spatial envelope, or whether the minimum distance between the position coordinates of the unassigned piece and the boundary of the spatial envelope is less than a preset envelope boundary distance threshold, wherein the preset envelope boundary distance threshold is determined based on the end effector end radius; S223: If the minimum distance between any two pieces in the initial group does not exceed the maximum opening and closing distance of the end effector, and the spatial envelope does not contain any other pieces not assigned to the initial group, and the minimum distance between the boundary of the spatial envelope and any unassigned piece is not less than the preset envelope boundary distance threshold, then the initial group is determined to meet the condition for parallel capture within the same group; otherwise, the initial group is determined not to meet the condition for parallel capture within the same group.

4. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 2, characterized in that, Step S23, which involves performing boundary adjustment operations on initial groups that do not meet the conditions for parallel crawling within the same group, includes: S231: For an initial group that does not meet the condition that the same group can be grabbed in parallel, determine that the area enclosed by the minimum bounding polygon contains unassigned pieces within the group, and remove the pieces within the group from the initial group. S232: For unassigned pieces that are spatially continuous with the initial group and can be captured by the same end effector, calculate the minimum distance between the unassigned pieces and the remaining pieces in the initial group. If the minimum distance does not exceed the maximum opening and closing distance of the end effector, then the unassigned pieces are incorporated into the initial group. S233: Repeat S231 and S232 until the initial group meets the condition that the same group can be crawled in parallel.

5. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 2, characterized in that, Step S24, which involves calculating the end effector interference probability between any two groups, includes: S241: For any two groups, obtain the three-dimensional motion trajectory of the end effector corresponding to all pieces in the two groups within a preset grasping time window, and perform a Cartesian product operation on the projection area of ​​the three-dimensional motion trajectory of each end effector on the chessboard plane and the preset grasping time window to obtain the spatiotemporal occupancy area of ​​the end effector. The preset grasping time window is the time interval for all end effectors in the same collaborative grasping group to complete one grasping operation. S242: Calculate the degree of overlap of the spatiotemporal occupied regions of all end effectors corresponding to the two groups to obtain the overlap parameter; calculate the ratio of the overlap parameter to the total volume of all spatiotemporal occupied regions corresponding to the two groups to obtain the end effector interference probability between the two groups. S243: If the end effector interference probability exceeds the preset interference threshold, it is determined that there is interference between the two groups. The pieces in the two groups are first redistributed. If the end effector interference probability between the two groups still exceeds the preset interference threshold after redistribution, the two groups are merged.

6. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 5, characterized in that, Step S241, which involves performing a Cartesian product operation between the projection area of ​​the three-dimensional motion trajectory of each end effector on the chessboard plane and the preset grasping time window to obtain the spatiotemporal occupancy area of ​​the end effector, includes: S2411: For each end effector, sample along the three-dimensional motion trajectory of the end effector within a preset grasping time window at a preset sampling interval to obtain multiple trajectory sampling points. The preset sampling interval is determined based on the ratio of the chess piece diameter to the motion speed of the end effector. S2412: Project each trajectory sampling point vertically onto the chessboard plane to obtain a set of planar projection points. Construct a minimum bounding polygon for the set of planar projection points and use the minimum bounding polygon as the projection area of ​​the end effector on the chessboard plane. S2413: Perform a Cartesian product operation on the projected area and the preset capture time window to obtain the spatiotemporal occupancy area of ​​the end effector. The spatiotemporal occupancy area is a prismatic spatiotemporal region with the projected area as the base and the preset capture time window as the height.

7. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 3, characterized in that, The preset envelope boundary distance threshold mentioned in step S222 is determined based on the end effector's end radius, including: S2221: Obtain the end shape parameters of the end effector and the cross-sectional shape parameters of the chess piece, and determine the shape compensation coefficient based on the end shape parameters of the end effector and the cross-sectional shape parameters of the chess piece. The shape compensation coefficient is used to characterize the degree of geometric adaptation when the end of the end effector contacts the edge of the chess piece. S2222: Obtain the number of pieces in the initial group, obtain the number of pieces per unit area of ​​the chessboard plane, wherein the number of pieces per unit area of ​​the chessboard plane is the ratio of the total number of remaining pieces on the chessboard to the chessboard plane area, obtain a preset reference area, wherein the preset reference area is a pre-set area benchmark value, and determine a density compensation coefficient based on the ratio of the number of pieces in the initial group to the number of pieces per unit area of ​​the chessboard plane and the ratio of the preset reference area, wherein the density compensation coefficient is used to characterize the additional influence of the density of pieces in the initial group on the safety margin of the spatial envelope boundary; S2223: Multiply the end radius of the end effector, the shape compensation coefficient, and the density compensation coefficient, and use the product as the preset envelope boundary distance threshold.

8. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 1, characterized in that, Step S3, which involves generating multi-end-efficiency actuator collaborative grasping instructions for each collaborative grasping group that match the number and spatial distribution of pieces within the collaborative grasping group, includes: S31: For each collaborative grasping group, obtain the static friction coefficient between each piece and the chessboard surface, the mass attribute of each piece, and the position of the piece assigned to be grasped by each end effector on the chessboard. S32: Calculate the micro-vibration sensitivity of each piece based on the static friction coefficient and the mass property. The micro-vibration sensitivity is used to characterize the sensitivity of the piece to positional displacement due to vibration transmission from the chessboard surface when other pieces in the same cooperative grasping group are contacted and grasped by the end effector. The micro-vibration sensitivity is negatively correlated with the static friction coefficient and is a dimensionless coefficient. S33: For any two pieces in the collaborative grasping group, calculate the vibration transmission coupling coefficient between the two pieces based on the minimum distance between them and the micro-vibration sensitivity, and construct a vibration transmission coupling matrix. Each element in the vibration transmission coupling matrix corresponds to the vibration transmission coupling coefficient between a pair of pieces, and the diagonal elements of the vibration transmission coupling matrix are zero. S34: The vibration transmission coupling matrix is ​​written as a feedforward compensation parameter into the multi-end effector collaborative grasping instruction. The feedforward compensation parameter is a parameter used to adjust the grasping force timing of the end effector before the grasping operation is executed. The multi-end effector collaborative grasping instruction also defines the grasping order, grasping path, grasping force parameter, and grasping force application time sequence of each end effector. The grasping force parameter includes the grasping force amplitude. The grasping force application time sequence is determined according to the vibration transmission coupling matrix, so that there is a preset time difference between the starting time of the grasping force application of each end effector in the same collaborative grasping group. The preset time difference is used to stagger the propagation of the chessboard surface vibration peak caused by the grasping force application. The preset grasping time window is the time interval for all end effectors in the same collaborative grasping group to complete one grasping operation.

9. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 8, characterized in that, The preset time difference mentioned in step S34 is determined according to the following steps: S3411: Obtain the vibration propagation speed of the chessboard surface material, wherein the vibration propagation speed is used to characterize the speed at which vibration energy propagates along the chessboard plane when the chessboard surface is subjected to the grasping force of the end effector. S3412: For any two end effectors in the same collaborative grasping group, obtain the minimum distance value between the chess pieces corresponding to the two end effectors respectively, and determine the end effector with the earlier grasping force application start time as the end effector that applies the grasping force first, and determine the end effector with the later grasping force application start time as the end effector that applies the grasping force later. S3413: Divide the minimum spacing value by the vibration propagation speed to obtain the vibration propagation time, which is used to characterize the time required for the vibration peak to propagate from one chess piece position to another chess piece position; S3414: Set the preset time difference to a value greater than the vibration propagation time, so that when the vibration peak generated by the end effector that applies the gripping force first propagates to the position of the chess piece corresponding to the end effector that applies the gripping force later, the end effector that applies the gripping force later has not yet started to apply the gripping force.

10. The method for collecting chess pieces in a chess-playing interactive device based on multi-piece collaborative grasping according to claim 8, characterized in that, The grasping force application time series mentioned in step S34 is generated according to the following steps: S3421: Using the vibration transmission coupling matrix as input, calculate the vibration superposition amplitude at each chess piece position on the chessboard surface based on the vibration transmission coupling matrix and the grasping force amplitude in the grasping force parameters of all end effectors. The vibration superposition amplitude is the maximum value of the vibration displacement on the chessboard surface. S3422: Taking the minimum amplitude of the vibration superposition as the adjustment target, within the adjustable range of the starting time of the grasping force application of each end effector, the starting time of the grasping force application of each end effector is iteratively adjusted. The adjustable range is the time interval determined by the product of the micro-vibration sensitivity of the chess piece corresponding to the end effector and the preset grasping time window. The micro-vibration sensitivity is a dimensionless coefficient. S3423: After each iteration of adjustment, the vibration superposition amplitude is recalculated according to the vibration transmission coupling matrix, and the adjustment results that reduce the vibration superposition amplitude are retained until the vibration superposition amplitude no longer decreases. S3424: Determine the starting time of the gripping force application of each end effector after iterative adjustment as the gripping force application time sequence, and write it into the multi-end effector collaborative gripping instruction.

Images