Double-arm robot cooperative motion planning method and system for multi-machine cooperative co-fusion

Abstract

The invention discloses a two-arm robot cooperative motion planning method and system for multi-machine cooperative co-fusion, and the method comprises the steps: constructing a machine-machine-environment co-fusion collision information estimation model: carrying out the discretization of the whole reachable space of a two-arm robot, and obtaining the environment information through a depth camera; a gap estimation neural network model is adopted to obtain interaction information of the double-arm robot and the environment, meanwhile, interaction information of left and right mechanical arms is obtained, and synchronous collision detection between the robots and between the robot and the environment is achieved; on the basis of the machine-machine-environment co-fusion collision information estimation model, performing two-arm robot collaborative path planning to obtain an estimated feasible path; and based on the geometric collision checker, carrying out double-arm robot motion path safety check and path repair, and ensuring absolute safety of a planned path. The problems that an existing two-arm robot motion planning method is low in efficiency, and multi-robot collaborative operation is difficult to effectively achieve are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of dual-arm robot collaborative motion planning, and in particular relates to a dual-arm robot collaborative motion planning method and system for multi-machine collaboration and integration. Background Art

[0002] Compared with a single robot arm, a dual-arm robot is not a simple combination of two robot arms, but significantly expands the working mode of the system. The dual-arm robot has higher flexibility and rigidity, and can switch between closed-chain and open-chain states in structure according to different tasks. It has both the flexibility of a serial structure and the rigidity of a parallel structure, which improves the control performance of the system. With its excellent performance, the dual-arm robot has shown broad application prospects in the industrial field. In order to adapt to complex tasks, the two robot arms need to coordinate and cooperate in a specific spatial environment to complete the task, that is, multi-robot collaboration and integration. When the distance between the two robot arms is too close, their workspaces will have overlapping areas, which is called a common workspace. Under the open-chain structure, the working characteristics of the dual-arm robot include: 1) There are overlapping workspaces; 2) The task objects are dispersed. The overlap of the workspace means that the robot arms will face the risk of collision when moving in the common area, and the dispersion of the task objects means that each robot arm is not constrained by the spatial position or posture of the common task object when moving. Therefore, the dual-arm robot under the open-chain structure needs to meet the following constraints when realizing synchronous collision-free path planning: in addition to avoiding environmental obstacles, it also needs to avoid collisions with other robot arms. Therefore, each robot arm must not only avoid external obstacles during the planning process, but also avoid potential collisions with adjacent robots. In addition, the high-dimensional joint configuration space of the dual-arm robot also greatly increases the difficulty of dual-arm robot planning.

[0003] Existing methods, such as sampling-based motion planning algorithms and artificial potential field methods, have low planning efficiency and are difficult to solve the problem of synchronous motion planning of dual-arm robots, and are still unable to effectively achieve multi-machine collaboration and integration. In response to the above problems and challenges, the present invention proposes a dual-arm robot collaborative motion planning method and system for multi-machine collaboration and integration. Summary of the invention

[0004] In response to the above technical problems, the present invention provides a dual-arm robot collaborative motion planning method and system for multi-machine collaboration and integration.

[0005] The technical solution adopted by the present invention to solve the technical problem is: A dual-arm robot collaborative motion planning method for multi-machine collaboration and integration, the method comprising the following steps: S100: setting the overall reachable space of the dual-arm robot and discretizing the reachable space; S200: Use the depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , respectively construct the left robotic arm occupancy matrix and the right robotic arm occupancy matrix ,in, and The element value setting judgment basis information is obtained based on the gap reasoning neural network model; S300: Occupancy matrix based on environmental obstacles , the left robotic arm occupies the matrix and the right robotic arm occupancy matrix Construct a machine-machine-environment co-integrated collision information estimation model; S400: Construct the starting configuration node and target configuration node of the dual-arm robot, initialize a path search tree, and perform batch random sampling in the dual-arm joint space Group configuration points and find the closest point to each sampling point in the path search tree path nodes, forming candidate path edges; S500: Perform equidistant interpolation on each edge to obtain interpolation points, and the front of each interpolation point Joint angle and posterior The joint angles of the left robot arm and the right robot arm are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm, and the machine-machine-environment fusion collision information estimation model is obtained by batch input. The collision estimation results of the interpolation points are calculated, and the last non-collision interpolation point among the interpolation points of each candidate path is selected as a new path extension node and added to the path search tree; S600: When the preset path planning end condition is met, the path planning process ends, and the dual-arm robot path check and safety repair are performed based on the geometric collision checker, and finally an effective absolutely collision-free safety path is obtained.

[0006] Preferably, S100 specifically includes: S110: Assume that the reachable spaces of the left and right arms of the dual-arm robot are represented as and ,right and Take the union to obtain the overall reachable space of the dual-arm robot ; S120: Yes and Take the intersection to obtain the common working space of the dual-arm robot ; S130: The overall reachable space of the dual-arm robot Split into Each voxel represents a size of The cubic space, The value is set by the user.

[0007] Preferably, S200 is specifically: S210: Use a depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , for dimensional matrix, where In The rules for setting element values ​​are as follows: If there is a point cloud in a voxel, it means there is an obstacle in the voxel. Corresponding to The element is set to 1, otherwise it is set to 0, where ; S220: Construct the left robot arm occupancy matrix separately and the right robotic arm occupancy matrix , both dimensional matrix, and The setting rules are as follows: Different from the environmental obstacles occupying the matrix , and The element value setting decision basis information is obtained based on the gap reasoning neural network model.

[0008] Preferably, in S220 and The element value setting judgment basis information is obtained based on the gap reasoning neural network model, specifically: S221: For the left and right robotic arms, the joint space of the left robotic arm is represented as , the joint space of the right robot arm is represented as ,in represents the degrees of freedom of the left and right robotic arms, assuming that the degrees of freedom of the left and right robotic arms are the same; S222: By configuring the joints of the left and right robotic arms and By inputting the gap estimation neural network model, the overall reachable space of the left and right robotic arms to the dual-arm robot can be obtained simultaneously. Inside The corresponding gap value matrix of the voxels and ,in, and Both dimensional matrix, and Both Dimensional matrix; S223: For the matrix or The elements, among which =1,2,…, , if its value is less than the safety threshold , indicating that the left or right robotic arm is in the accessible space Neidi If there is an intersection of the voxels, then the matrix or Neidi The element value is set to 1, otherwise it is set to 0. , .

[0009] Preferably, S300 includes: S310: Define the collision information matrix between the robot and the environment =( , if the matrix If there is an element with a value of 1 in , it indicates that the dual-arm robot collides with the environmental obstacle; otherwise, it indicates that the dual-arm robot does not collide with the environmental obstacle; S320: Define the machine-machine collision information matrix = ,if The existence of an element with a value of 1 indicates that there is a common workspace for the dual-arm robot If the left and right robotic arms collide with each other, it indicates that there is no collision between the left and right robotic arms.

[0010] Preferably, S400 includes: S410: Assume the starting configuration node of the left robotic arm of the dual-arm robot and the target configuration node In the joint space, they are expressed as and ; Starting configuration node for the right robot arm and the target configuration node In the joint space, they are expressed as and ; Treat the dual-arm robot as a whole for motion planning, that is, to build the starting configuration node of the dual-arm robot for , and the target configuration node for , initialize a path search tree , The initial nodes are ; S420: Batch random sampling in the joint space of both arms Group Configuration Point( ), among which Sampling point , Indicates The first sampling point of the robot arm joint angles, ; S430: For the above Group configuration points, find the path tree The closest to each sampling point Path Node ,form Candidate path edges.

[0011] Preferably, S500 includes: S510: Perform equidistant interpolation on each edge with an interpolation step length of , thus obtaining a total of Interpolation points ; S520: For Interpolation points , the front of each interpolation point Joint angle and posterior The joint angles of the left and right robot arms are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm. The machine-machine-environment fusion collision information estimation model can be obtained by batch inputting Collision estimation results of interpolation points; S530: Select the last non-collision interpolation point among the interpolation points on each candidate path as a new path extension node, and obtain a total of A new extension node is added to the path search tree. ,in, .

[0012] Preferably, the path planning end condition preset in S600 is specifically: The latest extension node satisfy , which indicates that an estimated feasible dual-arm robot path has been successfully found ,in, , preset by the user; or The path planning time limit has been reached.

[0013] Preferably, in S600, the dual-arm robot path check and safety repair are performed based on the geometric collision checker, and finally an effective absolutely collision-free safety path is obtained, including: Using the Geometry Collision Checker Is it possible for each path node of to collide? No collision, the dual-arm robot executes the motion path ; If any collision is detected, the following path safety repair process is initiated: The geometry collision checker identifies the colliding nodes ,in , is the number of collision nodes. For each collision point , select the nearest non-collision node before and after it as the local starting configuration node and local target node ,Then a path is planned between these local starting nodes using ,traditional path planning methods; If the initial repair attempt fails, the local starting configuration node and local target configuration nodes Adjust to be closer to the global starting configuration node and the global start configuration node target The above process is repeated until a valid absolutely collision-free safe path is found.

[0014] The dual-arm robot collaborative motion planning system for multi-machine collaboration and integration includes a reachable space discretization module, a robot arm occupancy matrix construction module, a machine-machine-environment integration collision information estimation model construction module, a candidate path edge determination module, a path search tree expansion module and a collision-free safe path determination module. The reachable space discretization module is used to set the overall reachable space of the dual-arm robot and discretize the reachable space; The robot arm occupancy matrix construction module is used to use the depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , respectively construct the left robotic arm occupancy matrix and the right robotic arm occupancy matrix ,in, and The element value setting judgment basis information is obtained based on the gap reasoning neural network model; Machine-machine-environment fusion collision information estimation model building module, used to calculate the collision information based on the environmental obstacle occupancy matrix. , the left robotic arm occupies the matrix and the right robotic arm occupancy matrix Construct a machine-machine-environment co-integrated collision information estimation model; The candidate path edge determination module is used to construct the starting configuration node and target configuration node of the dual-arm robot, initialize a path search tree, and perform batch random sampling in the dual-arm joint space. Group configuration points and find the closest point to each sampling point in the path search tree path nodes, forming candidate path edges; The path search tree extension module is used to perform equidistant interpolation on each edge to obtain interpolation points, and the front of each interpolation point Joint angle and posterior The joint angles of the left robot arm and the right robot arm are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm, and the machine-machine-environment fusion collision information estimation model is obtained by batch input. The collision estimation results of the interpolation points are calculated, and the last non-collision interpolation point among the interpolation points of each candidate path is selected as a new path extension node and added to the path search tree; The collision-free safe path determination module is used to end the path planning process when the preset path planning end conditions are met, and perform path checking and safety repair of the dual-arm robot based on the geometric collision checker, and finally obtain an effective absolutely collision-free safe path.

[0015] The above-mentioned dual-arm robot collaborative motion planning method and system for multi-machine collaboration and integration realizes synchronous collision detection between robots and robots, and between robots and the environment by constructing a machine-machine-environment integration collision information estimation model; in addition, based on the machine-machine-environment integration information model, a dual-arm robot collaborative planning strategy and path safety repair steps are designed, which is expected to ultimately realize multi-robot collaborative and integrated operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of a dual-arm robot collaborative motion planning method for multi-machine collaboration and integration in one embodiment of the present invention; Figure 2 Schematic diagram of path repair of a dual-arm robot in one embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings.

[0018] In one embodiment, Figure 1 As shown, a dual-arm robot collaborative motion planning method for multi-machine collaboration and integration, the method comprises the following steps: S100: setting the overall reachable space of the dual-arm robot and discretizing the reachable space; S200: Use the depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , respectively construct the left robotic arm occupancy matrix and the right robotic arm occupancy matrix ,in, and The element value setting judgment basis information is obtained based on the gap reasoning neural network model; S300: Occupancy matrix based on environmental obstacles , the left robotic arm occupies the matrix and the right robotic arm occupancy matrix Construct a machine-machine-environment co-integrated collision information estimation model; S400: Construct the starting configuration node and target configuration node of the dual-arm robot, initialize a path search tree, and perform batch random sampling in the dual-arm joint space Group configuration points and find the closest point to each sampling point in the path search tree path nodes, forming candidate path edges; S500: Perform equidistant interpolation on each edge to obtain interpolation points, and the front of each interpolation point Joint angle and posterior The joint angles of the left robot arm and the right robot arm are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm, and the machine-machine-environment fusion collision information estimation model is obtained by batch input. The collision estimation results of the interpolation points are calculated, and the last non-collision interpolation point among the interpolation points of each candidate path is selected as a new path extension node and added to the path search tree; S600: When the preset path planning end condition is met, the path planning process ends, and the dual-arm robot path check and safety repair are performed based on the geometric collision checker, and finally an effective absolutely collision-free safety path is obtained.

[0019] In one embodiment, S100 specifically includes: S110: Assume that the reachable spaces of the left and right arms of the dual-arm robot are represented as and ,right and Take the union to obtain the overall reachable space of the dual-arm robot ; S120: Yes and Take the intersection to obtain the common working space of the dual-arm robot ; S130: The overall reachable space of the dual-arm robot Split into Each voxel represents a size of The cubic space, The value is set by the user.

[0020] In one embodiment, S200 specifically includes: S210: Use a depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , for dimensional matrix, where In The rules for setting element values ​​are as follows: If there is a point cloud in a voxel, it means there is an obstacle in the voxel. Corresponding to The element is set to 1, otherwise it is set to 0, where ; S220: Construct the left robot arm occupancy matrix separately and the right robotic arm occupancy matrix , both dimensional matrix, and The setting rules are as follows: Different from the environmental obstacles occupying the matrix , and The element value setting judgment basis information (whether there is an intersection between the robotic arm and the voxel) does not come from the depth camera, but is obtained based on the gap reasoning neural network model.

[0021] In one embodiment, in S220 and The element value setting judgment basis information is obtained based on the gap reasoning neural network model, specifically: S221: For the left and right robotic arms, the joint space of the left robotic arm is represented as , the joint space of the right robot arm is represented as ,in represents the degrees of freedom of the left and right robotic arms, assuming that the degrees of freedom of the left and right robotic arms are the same; S222: By configuring the joints of the left and right robotic arms and By inputting the gap estimation neural network model, the overall reachable space of the left and right robotic arms to the dual-arm robot can be obtained simultaneously. Inside The corresponding gap value matrix of the voxels and ,in, and Both dimensional matrix, and Both Dimensional matrix; S223: For the matrix or The elements, among which =1,2,…, , if its value is less than the safety threshold ( , which can be set by the user), indicating that the left or right robotic arm is in the accessible space Neidi If there is an intersection of the voxels, then the matrix or Neidi The element value is set to 1, otherwise it is set to 0. , .

[0022] In one embodiment, the gap estimation neural network model includes four parts that are forward connected in sequence, as follows: The first part takes the robot joint configuration node as the initial input and uses the kernel function Encode the joint angles and increase the frequency of the input values, where Represents the input robot joint configuration node, It is an overclocking parameter; The second part consists of a fully connected layer, an activation function Relu function, and a DropOut layer. The output of the second part will be used as the input of the third part. The third part consists of a fully connected layer, an activation function Relu function, and a DropOut layer. The output of the third part will be used as the input of the fourth part. The fourth part consists of a fully connected layer, and the output feature number is , the purpose is to achieve the robot arm joint space to dimensional gap distance mapping.

[0023] In one embodiment, S300 includes: S310: Define the collision information matrix between the robot and the environment =( , if the matrix If there is an element with a value of 1 in , it indicates that the dual-arm robot collides with the environmental obstacle; otherwise, it indicates that the dual-arm robot does not collide with the environmental obstacle; S320: Define the machine-machine collision information matrix = ,if The existence of an element with a value of 1 indicates that there is a common workspace for the dual-arm robot If the left and right robotic arms collide with each other, it indicates that there is no collision between the left and right robotic arms.

[0024] The dual-arm robot needs to always ensure that and All of them are zero matrices, which means that all robotic arms are prevented from colliding with environmental obstacles and the left and right robotic arms are prevented from colliding with each other.

[0025] In one embodiment, S400 includes: S410: Assume the starting configuration node of the left robotic arm of the dual-arm robot and the target configuration node In the joint space, they are expressed as and ; Starting configuration node for the right robot arm and the target configuration node In the joint space, they are expressed as and ; Treat the dual-arm robot as a whole for motion planning, that is, to build the starting configuration node of the dual-arm robot for , and the target configuration node for , initialize a path search tree , The initial nodes are ; S420: Batch random sampling in the joint space of both arms Group Configuration Point( ), among which Sampling point , Indicates The first sampling point of the robot arm joint angles, ; S430: For the above Group configuration points, find the path tree The closest to each sampling point Path Node ,form Candidate path edges.

[0026] In one embodiment, S500 includes: S510: Perform equidistant interpolation on each edge with an interpolation step length of , thus obtaining a total of Interpolation points ; S520: For Interpolation points , the front of each interpolation point Joint angle and posterior The joint angles of the left and right robot arms are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm. The machine-machine-environment fusion collision information estimation model can be obtained by batch inputting Collision estimation results of interpolation points; S530: Select the last non-collision interpolation point among the interpolation points on each candidate path as a new path extension node, and obtain a total of A new extension node is added to the path search tree. ,in, .

[0027] In one embodiment, the path planning end condition preset in S600 is specifically: The latest extension node satisfy , which indicates that an estimated feasible dual-arm robot path has been successfully found ,in, , preset by the user; or the path planning time reaches the upper limit.

[0028] Planned in advance , due to the use of a machine-machine-environment symbiotic collision information estimation model containing a gap neural network, it is necessary to perform a safe repair of the path. This is because the neural network may make prediction errors, such as misestimating a robot arm configuration point that may collide as a collision-free result. Therefore, after successfully finding an estimated feasible dual-arm robot path After that, the dual-arm robot path inspection and safety repair are required.

[0029] In one embodiment, Figure 2 As shown in the figure, S600 performs dual-arm robot path checking and safety repair based on the geometric collision checker, and finally obtains an effective absolutely collision-free safety path, including: Using the Geometry Collision Checker Is it possible for each path node of to collide? No collision, the dual-arm robot executes the motion path ; If any collision is detected, the following path safety repair process is initiated: The geometry collision checker identifies the colliding nodes ,in , is the number of collision nodes. For each collision point , select the nearest non-collision node before and after it as the local starting configuration node and local target node ,Then a path is planned between these local starting nodes using the traditional path planning method (RRT or PRM method); If the initial repair attempt fails, the local starting configuration node and local target configuration nodes Adjust to be closer to the global starting configuration node and the global start configuration node target The above process is repeated until a valid absolutely collision-free safe path is found.

[0030] A dual-arm robot collaborative motion planning system for multi-machine collaboration and integration includes a reachable space discretization module, a robot arm occupancy matrix construction module, a machine-machine-environment integration collision information estimation model establishment module, a candidate path edge determination module, a path search tree expansion module, and a collision-free safe path determination module. The reachable space discretization module is used to set the overall reachable space of the dual-arm robot and discretize the reachable space; The robot arm occupancy matrix construction module is used to use the depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , respectively construct the left robotic arm occupancy matrix and the right robotic arm occupancy matrix ,in, and The element value setting judgment basis information is obtained based on the gap reasoning neural network model; The machine-machine-environment fusion collision information estimation model establishment module is used to calculate the collision information based on the environmental obstacle occupancy matrix. , the left robotic arm occupies the matrix and the right robotic arm occupancy matrix Construct a machine-machine-environment co-integrated collision information estimation model; The candidate path edge determination module is used to construct the starting configuration node and target configuration node of the dual-arm robot, initialize a path search tree, and randomly sample in batches in the dual-arm joint space. Group configuration points and find the closest point to each sampling point in the path search tree path nodes, forming candidate path edges; The path search tree extension module is used to perform equidistant interpolation on each edge to obtain interpolation points, and the front of each interpolation point Joint angle and posterior The joint angles of the left robot arm and the right robot arm are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm, and the machine-machine-environment fusion collision information estimation model is obtained by batch input. The collision estimation results of the interpolation points are calculated, and the last non-collision interpolation point among the interpolation points of each candidate path is selected as a new path extension node and added to the path search tree; The collision-free safe path determination module is used to end the path planning process when the preset path planning end conditions are met. The dual-arm robot path check and safety repair are performed based on the geometric collision checker, and finally an effective absolutely collision-free safe path is obtained.

[0031] For the specific definition of a dual-arm robot collaborative motion planning system for multi-machine collaboration and integration, please refer to the definition of a dual-arm robot collaborative motion planning method for multi-machine collaboration and integration mentioned above, which will not be repeated here. Each module in the above-mentioned dual-arm robot collaborative motion planning system for multi-machine collaboration and integration can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0032] The above dual-arm robot collaborative motion planning method and system for multi-machine collaboration and integration have the following technical effects: (1) The present invention innovatively constructs a machine-machine-environment integrated collision information estimation model, which can simultaneously realize efficient collision detection between robots and environmental obstacles, and between robots; (2) The present invention greatly simplifies the motion planning process of the dual-arm robot by considering the dual-arm robot as a whole for motion planning, thereby achieving efficient collaborative planning of the dual-arm robot; (3) The present invention not only utilizes the batch data processing capability of neural networks to achieve efficient machine-machine-environment collision detection, but also proposes corresponding motion path safety repair steps. While improving the planning efficiency, it also ensures the safety of the dual-arm robot during movement.

[0033] The above is a detailed introduction to the dual-arm robot collaborative motion planning method and system for multi-machine collaboration and integration provided by the present invention. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the core idea of ​​the present invention. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, the present invention can also be improved and modified, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A dual-arm robot collaborative motion planning method for multi-machine collaboration, characterized in that: The method comprises the following steps: S100: setting the overall reachable space of the dual-arm robot and discretizing the reachable space; S200: Use the depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , respectively construct the left robotic arm occupancy matrix and the right robotic arm occupancy matrix ,in, and The element value setting judgment basis information is obtained based on the gap reasoning neural network model; S300: Occupancy matrix based on environmental obstacles , the left robotic arm occupies the matrix and the right robotic arm occupancy matrix Construct a machine-machine-environment co-integrated collision information estimation model; S400: Construct the starting configuration node and target configuration node of the dual-arm robot, initialize a path search tree, and perform batch random sampling in the dual-arm joint space Group configuration points and find the closest point to each sampling point in the path search tree path nodes, forming candidate path edges; S500: Perform equidistant interpolation on each edge to obtain interpolation points, and the front of each interpolation point Joint angle and posterior The joint angles of the left robot arm and the right robot arm are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm, and the machine-machine-environment fusion collision information estimation model is obtained by batch input. The collision estimation results of the interpolation points are calculated, and the last non-collision interpolation point among the interpolation points of each candidate path is selected as a new path extension node and added to the path search tree; S600: When the preset path planning end condition is met, the path planning process ends, and the dual-arm robot path check and safety repair are performed based on the geometric collision checker, and finally an effective absolutely collision-free safety path is obtained.

2. The method according to claim 1, characterized in that S100 is specifically: S110: Assume that the reachable spaces of the left and right arms of the dual-arm robot are represented as and ,right and Take the union to obtain the overall reachable space of the dual-arm robot ; S120: Yes and Take the intersection to obtain the common working space of the dual-arm robot ; S130: The overall reachable space of the dual-arm robot Split into Each voxel represents a size of The cubic space, The value is set by the user.

3. The method according to claim 2, characterized in that S200 is specifically: S210: Use a depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , for dimensional matrix, where In The rules for setting element values ​​are as follows: If there is a point cloud in a voxel, it means there is an obstacle in the voxel. Corresponding to The element is set to 1, otherwise it is set to 0, where ; S220: Construct the left robot arm occupancy matrix separately and the right robotic arm occupancy matrix , both dimensional matrix, and The setting rules are as follows: Different from the environmental obstacles occupying the matrix , and The element value setting decision basis information is obtained based on the gap reasoning neural network model.

4. The method according to claim 3, characterized in that S220 and The element value setting judgment basis information is obtained based on the gap reasoning neural network model, specifically: S221: For the left and right robotic arms, the joint space of the left robotic arm is represented as , the joint space of the right robot arm is represented as ,in represents the degrees of freedom of the left and right robotic arms, assuming that the degrees of freedom of the left and right robotic arms are the same; S222: By configuring the joints of the left and right robotic arms and By inputting the gap estimation neural network model, the overall reachable space of the left and right robotic arms to the dual-arm robot can be obtained simultaneously. Inside The corresponding gap value matrix of the voxels and ,in, and Both dimensional matrix, and Both Dimensional matrix; S223: For the matrix or The elements, among which =1,2,…, , if its value is less than the safety threshold , indicating that the left or right robotic arm is in the accessible space Neidi If there is an intersection of the voxels, then the matrix or Neidi The element value is set to 1, otherwise it is set to 0. , .

5. The method according to claim 4, characterized in that S300 includes: S310: Define the collision information matrix between the robot and the environment =( , if the matrix If there is an element with a value of 1 in , it indicates that the dual-arm robot collides with the environmental obstacle; otherwise, it indicates that the dual-arm robot does not collide with the environmental obstacle; S320: Define the machine-machine collision information matrix = ,if The existence of an element with a value of 1 indicates that there is a common workspace for the dual-arm robot If the left and right robotic arms collide with each other, it indicates that there is no collision between the left and right robotic arms.

6. The method according to claim 5, characterized in that S400 includes: S410: Assume the starting configuration node of the left robotic arm of the dual-arm robot and the target configuration node In the joint space, they are expressed as and ; Starting configuration node for the right robot arm and the target configuration node In the joint space, they are expressed as and ; Treat the dual-arm robot as a whole for motion planning, that is, to build the starting configuration node of the dual-arm robot for , and the target configuration node for , initialize a path search tree , The initial nodes are ; S420: Batch random sampling in the joint space of both arms Group Configuration Point( ), among which Sampling point , Indicates The first sampling point of the robot arm joint angles, ; S430: For the above Group configuration points, find the path tree The closest to each sampling point Path Node ,form Candidate path edges.

7. The method according to claim 6, characterized in that S500 includes: S510: Perform equidistant interpolation on each edge with an interpolation step length of , thus obtaining a total of Interpolation points ; S520: For Interpolation points , the front of each interpolation point Joint angle and posterior The joint angles of the left and right robot arms are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm. The machine-machine-environment fusion collision information estimation model can be obtained by batch inputting Collision estimation results of interpolation points; S530: Select the last non-collision interpolation point among the interpolation points on each candidate path as a new path extension node, and obtain a total of A new extension node is added to the path search tree. ,in, .

8. The method according to claim 7, characterized in that The path planning end condition preset in S600 is specifically: The latest extension node satisfy , which indicates that an estimated feasible dual-arm robot path has been successfully found ,in, , preset by the user; or The path planning time limit has been reached.

9. The method according to claim 8, characterized in that In S600, the dual-arm robot path check and safety repair are performed based on the geometric collision checker, and an effective absolutely collision-free safety path is finally obtained, including: Using the Geometry Collision Checker Is it possible for each path node of to collide? No collision, the dual-arm robot executes the motion path ; If any collision is detected, the following path safety repair process is initiated: The geometry collision checker identifies the colliding nodes ,in , is the number of collision nodes. For each collision point , select the nearest non-collision node before and after it as the local starting configuration node and local target node ,Then a path is planned between these local starting nodes using ,traditional path planning methods; If the initial repair attempt fails, the local starting configuration node and local target configuration nodes Adjust to be closer to the global starting configuration node and the global start configuration node target The above process is repeated until a valid absolutely collision-free safe path is found.

10. A dual-arm robot collaborative motion planning system for multi-machine collaboration, characterized in that: It includes a reachable space discretization module, a robot occupancy matrix construction module, a machine-machine-environment co-integration collision information estimation model construction module, a candidate path edge determination module, a path search tree expansion module and a collision-free safe path determination module. The reachable space discretization module is used to set the overall reachable space of the dual-arm robot and discretize the reachable space; The robot arm occupancy matrix construction module is used to use the depth camera to obtain the obstacle point cloud information within the overall reachable space of the dual-arm robot and construct the environmental obstacle occupancy matrix , respectively construct the left robotic arm occupancy matrix and the right robotic arm occupancy matrix ,in, and The element value setting judgment basis information is obtained based on the gap reasoning neural network model; The machine-machine-environment fusion collision information estimation model building module is used to estimate the collision information based on the environmental obstacle occupancy matrix. , the left robotic arm occupies the matrix and the right robotic arm occupancy matrix Construct a machine-machine-environment co-integrated collision information estimation model; The candidate path edge determination module is used to construct the starting configuration node and target configuration node of the dual-arm robot, initialize a path search tree, and perform batch random sampling in the dual-arm joint space. Group configuration points and find the closest point to each sampling point in the path search tree path nodes, forming candidate path edges; The path search tree extension module is used to perform equidistant interpolation on each edge to obtain interpolation points, and the front of each interpolation point Joint angle and posterior The joint angles of the left robot arm and the right robot arm are respectively composed of the joint angle configuration points of the left robot arm and the right robot arm, and the machine-machine-environment fusion collision information estimation model is obtained by batch input. The collision estimation results of the interpolation points are calculated, and the last non-collision interpolation point among the interpolation points of each candidate path is selected as a new path extension node and added to the path search tree; The collision-free safe path determination module is used to end the path planning process when the preset path planning end conditions are met, and perform path checking and safety repair of the dual-arm robot based on the geometric collision checker, and finally obtain an effective absolutely collision-free safe path.

Images