Intelligent disassembly robot system and control method

Abstract

The present invention belongs to the field of robot automation control, and specifically is an intelligent disassembly robot system and control method, comprising: a robot working platform and a rectangular coordinate robot system, a dual-arm robot system, a mechanical gripping mechanism and a disassembly control system arranged on the robot working platform; a mechanical gripping mechanism is arranged between the rectangular coordinate robot system and the dual-arm robot system; the disassembly control system is respectively connected to the rectangular coordinate robot system, the dual-arm robot system and the mechanical gripping mechanism; the disassembly control system comprises an industrial computer, an electrical control panel and a PLC connected in sequence; a platform pan-tilt head and a camera are also arranged on the robot working platform; the beneficial effect of the present invention is that through a control box, the movement of the bomb disposal task is remotely and accurately controlled in a closed-loop motion control manner, and the operation status is monitored in real time through the depth camera and multiple cameras in the present invention, and the present invention can be widely used in various bomb disposal tasks of dismantling explosives.

Description

Technical Field

[0001] The invention belongs to the field of robot automation control, and in particular to an intelligent disassembly robot system and a control method. Background Art

[0002] Faced with suspicious packages disguised as "packages", existing dismantling robots are unable to quickly and effectively dispose of them. In addition, most existing dismantling robots rely entirely on remote control by personnel, and cannot be controlled and adjusted in real time according to actual environmental sensor information. The level of intelligence is not high, which can easily lead to burden and fatigue of operators. In particular, personnel need to concentrate highly, and the complexity and danger of the environment can easily lead to fatigue, making it impossible to ensure operational accuracy, which can easily lead to misoperation during the dismantling process. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent disassembly robot system and a control method to solve the technical problems in the above-mentioned background technology.

[0004] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: an intelligent disassembly robot system, including: a robot working platform and a rectangular coordinate robot system, a dual-arm robot system, a mechanical gripper mechanism and a disassembly control system arranged on the robot working platform;

[0005] A mechanical gripping mechanism is provided between the rectangular coordinate robot system and the dual-arm robot system;

[0006] The disassembly control system is respectively connected with the rectangular coordinate robot system, the dual-arm robot system, and the mechanical gripper mechanism;

[0007] The disassembly control system includes an industrial computer, an electrical control panel, and a PLC connected in sequence;

[0008] The PLC is connected to a plurality of servo drivers, and the plurality of servo drivers are respectively connected to the servo motors of the rectangular coordinate robot system, the dual-arm robot system, and the mechanical gripper mechanism to perform corresponding actions;

[0009] The electrical control panel is connected to the dual-arm robot system through a servo driver to control the action execution of the dual-arm robot system; the electrical control panel is connected to the rectangular coordinate robot system to receive the environmental information of the target to be disassembled collected by the rectangular coordinate robot system;

[0010] The electrical control board is used to receive control instructions sent by the industrial computer, control the dual-arm robot system to perform corresponding actions, and receive the control information of the motor obtained by the electrical control board after processing the control instructions and environmental information, and send it to the PLC to control the movement of the rectangular coordinate robot system and the mechanical gripper mechanism.

[0011] The robot operation platform is also provided with a platform pan / tilt and a camera;

[0012] There are two platform pan-tilt platforms, which are respectively arranged on one side vertically fixed on the robot working platform and perpendicular to the movement direction of the dual-arm robot system; each platform pan-tilt platform is provided with a camera that can realize the pitch function, and the camera is connected to the industrial computer through a video acquisition card.

[0013] The rectangular coordinate robot system includes: a disassembly tool box, an X-axis linear slide module, a cantilever Y-axis linear slide module, a Z-axis linear slide module and a depth camera;

[0014] The X-axis linear slide module is installed on one side of the robot working platform; the slide A of the X-axis linear slide module is fixedly connected to the Z-axis linear slide module, so that the Z-axis linear slide module is driven by the slide A to perform linear motion along the X-axis direction;

[0015] The base of the cantilever Y-axis linear slide module is installed on the slide B of the Z-axis linear slide module through an adapter plate, so that the cantilever Y-axis linear slide module can move linearly along the Z-axis through the slide B;

[0016] The disassembly tool box is fixed on the slide C of the cantilever Y-axis linear slide module;

[0017] Industrial cameras are symmetrically arranged on both sides of the disassembly toolbox, and the industrial cameras are connected to the industrial computer through a video acquisition card, and are used to transmit multiple video signals to the industrial computer in real time and display them on a display screen connected to the industrial computer;

[0018] The depth camera is fixed on the card seat below the disassembly tool box and connected to the electrical control board of the disassembly control system to collect environmental information of the target to be disassembled;

[0019] The servo motors corresponding to the X-axis linear slide module, the cantilever Y-axis linear slide module, and the Z-axis linear slide module are connected to the PLC.

[0020] A conversion plate is horizontally provided on the slide A of the X-axis linear slide module, and a vertical plate is vertically provided on the conversion plate. One side of the vertical plate is fixedly connected to the conversion plate, and the other side is fixedly connected to the base of the Z-axis linear slide module, so that the Z-axis linear slide module is perpendicular to the X-axis linear slide module and moves linearly along the X-axis direction with the slide A.

[0021] The dual-arm robot system includes: a pan-tilt head, a binocular camera, a seven-degree-of-freedom robotic arm, a two-degree-of-freedom trunk, a gripper, a cushion seat, a walking axis linear slide module and a slide seat D;

[0022] The bottom end of the two-degree-of-freedom trunk is fixed to the robot's pad by bolts, and the robot's pad is fixed to the slide D of the walking axis linear slide module. The walking axis linear slide module is arranged in the robot's working platform, and the slide D is arranged on the slideway on the surface of the robot's working platform. The movement direction of the walking axis linear slide module is parallel to the X-axis linear slide module, so that the slide D drives the two-degree-of-freedom trunk to move along the X-axis direction;

[0023] A pan-tilt is provided at the top of the two-degree-of-freedom trunk, and a binocular camera is provided on the pan-tilt. The binocular camera is connected to the input end of the control box industrial computer through a video acquisition card, and is used to transmit multiple video signals to the control box in real time, and display them on the display screen, so as to monitor and observe the disassembly work and provide guidance for the entire disassembly work; the output end of the control box is connected to the pan-tilt, and is used to adjust the posture of the binocular camera by controlling the pan-tilt;

[0024] Seven-degree-of-freedom mechanical arms are symmetrically arranged on both sides of the top of the two-degree-of-freedom trunk, and claws are arranged at the ends of the seven-degree-of-freedom mechanical arms;

[0025] A mechanical gripping mechanism is provided on the robot working platform between the seven-degree-of-freedom mechanical arm and the cantilever Y-axis linear slide module;

[0026] The servo motor of the walking axis linear slide module is connected to the corresponding servo driver in the control cabinet; the servo motors corresponding to the two-degree-of-freedom trunk and the seven-degree-of-freedom robotic arm are respectively connected to the electrical control board through the corresponding servo drivers.

[0027] The mechanical gripper mechanism comprises: a positive and negative tooth bidirectional linear slide module, a gripper base, a slide seat E, a gripper support frame, a rubber part, and a mechanical gripper mechanism;

[0028] The mechanical gripper mechanism is slidably mounted on the gripper support frame, and there are two mechanical grippers arranged opposite to each other. Rubber parts are arranged on opposite sides of the mechanical grippers to prevent the entire mechanical gripper from loosening when vibration and impact occur.

[0029] There are two gripper support frames, which are respectively threadedly connected to the corresponding gripper bases; the gripper bases are arranged on the corresponding positive and negative thread bidirectional linear slide module through a slide seat;

[0030] The servo motor of the forward and reverse bidirectional linear slide module is connected to the PLC through a corresponding servo driver.

[0031] A control method for an intelligent disassembly robot system comprises the following steps:

[0032] 1) The industrial computer controls the pan / tilt and camera of the dual-arm robot, and displays multiple video signals on the display screen in real time to monitor and observe the disassembly work. The industrial computer sends control instructions to the electrical control board, which receives environmental information fed back by the depth camera.

[0033] 2) The depth camera transmits the color image and the depth image in the environmental information of the target to be disassembled to the electrical control board, and the electrical control board processes the collected images to obtain the coordinates of the feature points of the target to be disassembled in the coordinate system of the robot working platform;

[0034] 3) The electrical control board plans the motion of the dual-arm robot system's manipulators based on the coordinates of the feature points of the target to be disassembled in the robot's work platform coordinate system, and obtains the planning path information using a fast-expanding random tree algorithm;

[0035] 4) The electrical control board obtains the rotation angles of each joint of the dual-arm robot system through inverse kinematics solution based on the planned path information, and converts the rotation angles into control information of the motor;

[0036] 5) The electrical control board controls the servo motor of the dual-arm robot through CAN bus communication to control the movement of the dual-arm robot and sends the control information of the motor to the PLC. The PLC controls the corresponding servo motor feed through the corresponding servo driver according to the control information of the motor to realize the control of the linear slide module.

[0037] In step 2), the electrical control board processes the collected image to obtain the coordinates of the target to be disassembled, including the following steps:

[0038] After the depth camera acquires the initial point cloud image;

[0039] (2.1) Processing the initial point cloud data through straight-through filtering to remove invalid point cloud data;

[0040] (2.2) The point cloud is sparsely made while ensuring the shape characteristics of the point cloud through the voxel filtering algorithm, so as to reduce the amount of point cloud data;

[0041] (2.3) Point cloud segmentation: Match and identify the platform in the point cloud data according to the trained plane model; input the sparse point cloud image into the plane model, extract the plane point cloud data subset on the platform according to the distribution of the platform point cloud data, and obtain the plane point cloud map;

[0042] The trained plane model is a model that recognizes a point cloud image of a plane from the entire environment image;

[0043] (2.4) Point cloud clustering: Obtain all point clouds on the plane point cloud map, and divide the point cloud data of the target to be disassembled on the plane point cloud map into multiple independent small point cloud sets; obtain the coordinates of the feature points of the target to be disassembled based on the camera depth coordinate system;

[0044] (2.5) Coordinate transformation: Perform coordinate transformation on the feature points of the target to be disassembled based on the camera depth coordinate system to obtain the coordinates of the feature points of the target object in the robot work platform coordinate system.

[0045] The step 3) comprises the following steps:

[0046] 3.1) Set the starting point X of the end of the robot arm init and the target point X at the end of the robot arm goat ,Add obstacle environment information;

[0047] The obstacle environment information is: point cloud data of an object that blocks the end of the robot arm in the plane point cloud map to represent the obstacle;

[0048] 3.2) Expand the leaf nodes by expanding the formula:

[0049] a) Determine dist(X new ,X goal ). If L≥dist(X new ,X goal ) then the expansion ends and this loop is exited;

[0050] b) If L<dist(X new ,X goal ) means that the planned path does not reach the end point from the starting point, then an extension node is obtained and T is traversed. k , find X new . Select X rand Expand the leaf nodes, that is, select 20 nodes as candidate random nodes X rand , calculate dist(X rand ,X goal ), save it to array

[20] , sort it from small to large, and select the first four smallest random nodes X through roulette wheel rand ;

[0051] Among them, the T k is a random tree with k nodes representing the positions of the robot end, X rand is a state point randomly selected in the Cartesian space C of the robot; L is the search step length, and dist represents the Euclidean distance

[0052] The search for X new , that is: find the distance X rand The nearest node Xnearst ; then from X nearst and X rand Find the distance X in nearst X rand and X nearst The node X that is closer to new ;

[0053] c) Enter 4 cycles, for each random node X rand , if dist(X new ,X rand )<L, then do not expand; if X is not found after completing 4 cycles rand , return to step a);

[0054] Judge X new Check whether the line connecting the node to its parent node touches any obstacle in the scene. If no collision occurs, expand the node and return to step a).

[0055] When the loop ends, it is considered that the expanded leaf node is found, and returns to step a);

[0056] 3.3) Use the robot kinematics to find the inverse solution for each pose point of each path point to obtain the angles of all joints;

[0057] 3.4) After the inverse solution, the angles of all joints are obtained, and the kinematic solution is performed to obtain the spatial coordinates corresponding to the joint angles, and collision detection is performed again;

[0058] 3.5) If the inverse solution fails, randomly select a value between the angles of two adjacent points in the posture until a joint angle that satisfies the safety of the robot arm in this posture is obtained.

[0059] The step 4) is specifically:

[0060] According to the planned path information, the Newton-Simpson iteration method is used to perform kinematics solution to obtain the joint angles of the robot arm. The specific steps are as follows:

[0061] 4.1) Using positive kinematics and the current angle value of a joint of the robot arm: q cur , get the position and posture of a joint at the end of the current robot arm, that is, the posture: T cur ;

[0062] 4.2) According to the posture of a joint target of the robot arm, obtain the differential motion matrix Δ of a joint of the robot arm;

[0063] 4.3) According to the differential operator in the current differential motion matrix Δ, the differential vector D of the robot end motion is obtained, and according to the current differential vector D, the differential motion of a joint of the robot arm is represented by the differential vector D; according to the inverse of the Jacobian matrix as the pseudo-inverse of the robot arm, the size of dq is calculated;

[0064] 4.4) According to the currently calculated dq, it is added to the current joint angle, that is, the joint angle changes to q cur =dq+q cur ;

[0065] 4.5) According to the q of the remaining joints of the robot arm cur Calculate the posture T of the remaining joint ends of the robot arm cur , return to 4.1) to continue iterating;

[0066] Set the termination condition for the iteration process. When the actual maximum accuracy error e meets the threshold, or the maximum number of iterations n reaches the set number of iterations N, the iteration is terminated.

[0067] The actual maximum accuracy error is:

[0068] The current end position of the robot arm T cur The terminal pose T with the desired target end , the difference between them, namely: ||T cur -T end ||<e.

[0069] The present invention has the following beneficial effects and advantages:

[0070] 1. The intelligent disassembly robot system of the present invention includes a rectangular coordinate robot and a dual-arm robot. The rectangular coordinate robot and its terminal disassembly tool can quickly complete disassembly actions that are difficult for the dual-arm robot to complete, such as large-scale moving and large-scale cutting. The dual-arm robot can complete two-handed coordinated disassembly actions that are difficult for the rectangular coordinate robot to complete, such as cutting wires, pulling wires, and other disassembly actions, as well as small-scale grasping and moving disassembly actions. The two robots are combined together with the help of an adjustable robot work platform, and with the mechanical gripping mechanism of the present invention, they can cooperate to complete various complex tasks that are difficult for other bomb disposal robots to complete, such as the two robots cooperate to move different disassembly objects to a suitable position, and then use the mechanical gripping mechanism to fix the disassembly objects, and then the two robots perform different disassembly actions to cooperate to complete different tasks, so as to better deal with various suspicious packages suspected of containing explosives. Therefore, the present invention has a high anti-terrorism bomb disposal application value.

[0071] 2. The multiple cameras or depth cameras of the present invention can provide good support for the robot bomb disposal operation: make full use of the color information and depth information of the image collected by the depth camera to locate the suspected explosives suspected of containing explosives, guide the rectangular coordinate robot and the dual-arm robot to approach the target and perform basic treatment such as cutting and clamping, etc., reduce the burden and fatigue of the bomb disposal operators, and provide support for subsequent fine operations; the two industrial cameras on the pan-tilt at the far end of the platform can achieve full-angle coverage of the entire robot on the work platform through precise manual adjustment of the adjustable components. The two cameras at the end of the rectangular robot can monitor, observe and guide the operator's fine operation of the rectangular robot in real time, and the pan-tilt and camera of the dual-arm robot can observe and guide the bomb disposal operator's fine operation of the dual-arm robot. It effectively improves the intelligence level of the bomb disposal operation, reduces the burden and fatigue of the bomb disposal operator, and reduces the misoperation caused by the fatigue of the operator during the bomb disposal operation, avoiding safety hazards.

[0072] 3. The present invention is also provided with a robot control box for remote control, which uses closed-loop motion control to achieve remote and precise control of the movement of the bomb disposal task, and monitors the operation status in real time through the visual system of the present invention, so that the bomb disposal task implementation process only requires one bomb disposal personnel who remotely controls the equipment to complete it. The present invention can be widely used in various bomb disposal tasks for dismantling explosives. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0074] Figure 2 It is a schematic diagram of the overall structure of the present invention;

[0075] Figure 3 It is a schematic diagram of the three-dimensional structure of the dual-arm robot and its walking axis unit in the present invention;

[0076] Figure 4 It is a schematic diagram of the overall structure of the mechanical gripper mechanism assembly of the present invention;

[0077] Figure 5 It is an exploded schematic diagram of the component structure of the mechanical gripping mechanism in the present invention;

[0078] Figure 6 is a schematic diagram of a robot control box in the present invention;

[0079] Figure 7 This is a schematic diagram of the motion flow of the vision-guided robot in the present invention;

[0080] Figure 8 It is a schematic diagram of the image processing algorithm flow in the present invention;

[0081] Fig. 9 It is a structural block diagram of the overall control system in the present invention;

[0082] Among them, 1 is a rectangular coordinate robot system, 101 is a slide B, 103 is a disassembly tool box, 104 is a camera connection component, 105 is an industrial camera, 106 is a disassembly tool, 107 is an X-axis linear slide module, 108 is a slide A, 109 is a cantilever Y-axis linear slide module, 110 is a Z-axis linear slide module, 111 is an adapter plate, 112 is a vertical plate, 113 is a reinforcing rib, 114 is a conversion plate, 115 is a slide C, 116 is a depth camera, 2 is a dual-arm robot system, 201 is a pan / tilt, 202 is a binocular camera, 203 is a seven-degree-of-freedom robot arm end, 204 is a two-degree-of-freedom torso, 205 is a gripper, 206 is a robot pad seat, 207 is a walking axis linear slide module, 208 is a slide D, and 3 is a machine Human working platform, 301 is disassembly working platform, 302 is linear slide module installation platform, 303 is electrical equipment placement platform, 304 is electrical control cabinet, 305 is universal caster, 306 is platform pan / tilt, 307 is camera, 308 is electrical control board, 4 is mechanical gripping mechanism, 401 is positive and negative tooth bidirectional slide module, 402 is gripping base, 403 is slide seat E, 404 is gripping support frame, 405 is rubber part, 406 is mechanical gripping mechanism, 408 is lead screw nut, 409 is positive and negative tooth ball screw, 410 is linear guide rail, 411 is baffle, 412 is aluminum alloy profile, 413 is servo motor, 414 is bearing seat, 415 is synchronous wheel, 416 is motor cover, 417 is photoelectric switch, 418 is guide rail slider. 5 is a control box, 501 is a display screen, 502 is a control panel, 503 is an upper box body, and 504 is a lower box body. DETAILED DESCRIPTION

[0083] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0084] Figure 1 , Figure 2 It is a schematic diagram of the overall structure of the present invention.

[0085] Figure 3 Three views of the cantilever structure of the rectangular coordinate robot of the present invention

[0086] Figure 4 Schematic diagram of the three-dimensional structure of the dual-arm robot and its walking axis unit in the present invention

[0087] like Figures 1 to 4As shown, the robot working platform in the present invention is composed of a platform frame 3 and universal casters 305. The platform frame is composed of square tubes through welding and has a three-layer structure, including the top disassembly work platform of the robot working platform 3, on which packages suspected of containing explosives are placed and disassembly operations are carried out, the second-layer linear slide module installation platform of the robot working platform 3, including the X-axis linear slide module of the rectangular coordinate robot system, the forward and reverse bidirectional linear slide module equipped with a mechanical gripper mechanism, and the linear slide module as the walking axis unit of the dual-arm robot; the third-layer electrical equipment placement platform of the robot working platform 3, including a control cabinet and an electrical control panel.

[0088] The disassembly operation platform 301 is fixed on the top layer of the robot operation platform 3, and is composed of a metal plate fixed on the robot operation platform 3 by threaded connection. The disassembly operation platform 301, the second layer of the robot operation platform 3 is a linear slide module installation platform 302, and the third layer of the robot operation platform 3 is an electrical equipment placement platform 303. The electrical control cabinet 304 and the electrical control board 308 in the present invention are installed on the electrical equipment placement platform 303. The pan-tilt 306 of the other two industrial cameras 307 is fixed at the far end of the disassembly operation platform 301, and the tube shell of the camera assembly 104 and the pan-tilt flange pipe fitting are clearance-matched and fixed with a top screw. The horizontal rotation and pitch movement of the two industrial cameras 307 are realized under the precise manual adjustment of the platform pan-tilt 306, so that the two cameras 307 can rotate arbitrarily in the horizontal direction, and the maximum pitch angle is ±45°, thereby realizing the adjustment of the azimuth and pitch angles of the two industrial cameras 307. The two industrial cameras 307 are responsible for monitoring and observing the working status of the entire disassembly operation from the far end of the platform.

[0089] The pan-tilt platform 201 and binocular camera 202 of the dual-arm robot system 2 are connected to the industrial computer of the robot control box 5 , and the posture of the pan-tilt platform 201 is adjusted under the control of the industrial computer of the control box 5 .

[0090] The intelligent disassembly robot system of the present invention comprises a rectangular coordinate robot system 1 and a dual-arm robot system 2 .

[0091] The linear module of the rectangular coordinate robot system 1 is composed of a ball screw, a linear guide, an aluminum alloy profile, a coupling, a servo motor, a photoelectric switch, etc., and is fixed inside the second layer of the platform frame through threaded connection and cylindrical pin interference fit.

[0092] The X-axis linear slide module 107 of the rectangular coordinate robot system 1 is installed on the robot work platform 3, and the Z-axis linear slide module 110 is slidably coupled to the slide 108 of the X-axis linear slide module through the conversion plate 114, the reinforcing ribs 113 and the vertical plate 112, so that it can perform linear motion along the X-axis direction. The cantilever Y-axis linear slide module 109 is installed on the slide 101 of the Z-axis linear slide module through the adapter plate 111, and can perform linear motion along the Z-axis through the slide 101. The disassembly tool box 103 is fixed on the slide 115 of the Y-axis linear slide module. The disassembly tool 106 in this example is an explosion-proof electric grinder held by the clamp and is fixed in the tool box 103 by bolts and elastic washers, and the depth camera 116 is fixed on the base of the tool box 103.

[0093] The disassembly tool 106 and the depth camera 116 are combined through the disassembly tool box 103, and can slide along the Y axis of the rectangular coordinate robot 1 through the slide 115 of the Y axis linear slide module 109, and with the help of the Y axis linear slide module 109 and the X axis linear slide module 107, perform linear motion along the X axis and the Y axis, so as to realize the three-dimensional motion of the end of the rectangular coordinate robot system 1 in space;

[0094] The depth camera 116 is a structured light depth camera realsensed435, which is powered by a ready-to-use USB and is equipped with a depth module, two infrared sensors, an infrared laser transmitter, and a color sensor. The depth camera 116 is installed on the connecting card seat of the disassembly tool box 103 of the slide C115 of the cantilever Y-axis linear slide module 109 of the rectangular coordinate robot system 1. It moves with the end of the rectangular coordinate robot system 1 to obtain the color information and depth information of the entire working environment and locate the position of the package.

[0095] Two industrial cameras 105 are installed on both sides of the disassembly tool box 103 through a camera connecting assembly 104. Each of the camera connecting assemblies 104 also includes a fine-tuning mechanism; the fine-tuning mechanism includes bolts, locking nuts and spring washers; wherein each of the bolts passes through a radial threaded hole on the tube shell of the camera connecting assembly 104, and is connected to the pipe fitting of the camera connecting assembly 104 through bolts, locking nuts and spring washers to complete the pitch movement of the industrial camera 105 to adjust the shooting angle and fix the camera. The maximum pitch angle observed by the two industrial cameras 105 is ±45°.

[0096] The dual-arm robot 2 is connected to the robot pad 206 by bolts, and the robot pad 206 is fixed to the slide 208 of the dual-arm robot walking axis linear slide module 207 by bolts and cylindrical pins. The walking axis linear slide module 207 is fixedly installed on the second-layer linear slide module installation platform 302 of the platform frame 3 by threaded connection and cylindrical pin interference fit. The dual-arm robot 2 is driven to walk by the sliding of the slide 208 of the walking axis linear slide module 207 on the linear slide module. The two seven-degree-of-freedom mechanical arms 203 of the dual-arm robot 2 have grippers 201 at the ends, and a trunk 204 with two degrees of freedom. At the same time, the head of the dual-arm robot has a pan-tilt 201 and a head camera 202. The posture of the camera 202 is adjusted by controlling the pan-tilt 201.

[0097] An intelligent disassembly robot and control system, including a robot system, a matching modular adjustable robot operation platform, a visual system and a control system. The robot system includes a rectangular coordinate robot and a dual-arm robot; the visual system includes a depth camera, an industrial camera, a pan / tilt and camera of a dual-arm robot, and a video acquisition card; the control system mainly includes an electrical control cabinet, an electrical control panel, a robot control box, etc.

[0098] The visual system mainly includes a depth camera 116, a pan / tilt 201 and a binocular camera 202 of a dual-arm robot system 2, four industrial cameras, and a video acquisition card;

[0099] Two industrial cameras 105 are hoisted on both sides of the toolbox at the end of the rectangular coordinate robot through the camera connection component 104. By adjusting the camera connection component 104, the pitch movement of the two industrial cameras can be achieved to monitor the disassembly work through real-time video observation.

[0100] Figure 5 It is a schematic diagram of the overall structure of the mechanical gripper mechanism assembly in the present invention,

[0101] Figure 6 Schematic diagram of the exploded structure of the mechanical gripper mechanism component in the present invention

[0102] like Figure 5 , Figure 6As shown, the mechanical gripper mechanism 406 and the gripper support frame 404 have mutually cooperating protrusions and slide grooves for detachable connection. The protrusion of the mechanical gripper mechanism 406 is placed in the slide groove and slidably cooperates with the slide groove, so that the mechanical gripper 406 can slide in the slide groove, while preventing the mechanical gripper 406 from shaking left and right, and is fixed in a suitable position by bolts and spring washers. The gripper support frame 404 is connected to the gripper base 402 through a threaded connection, and the gripper base 402 is installed on the positive and negative bidirectional linear slide module 401. The gripper base 402 is installed on the slide 403 of the positive and negative bidirectional linear slide module 401 through a threaded connection, and is fixed in the pin hole of the slide 403 through an elastic cylindrical pin interference fit, thereby realizing the precise positioning of the mechanical gripper mechanism 406 on the positive and negative bidirectional linear slide module 401, and preventing the loosening of the connection of the entire mechanical gripper assembly 4 device when vibration and impact occur. The two rubber parts 405 are adhered to the surfaces of the two mechanical gripper mechanisms 406 by high-strength bonding, thereby increasing the roughness of the contact surface of the mechanical gripper mechanism 406 when clamping to increase the friction, and providing flexible protection for the clamped package, reducing the pressure on the package during the clamping process, and preventing trigger damage to suspicious packages containing explosives.

[0103] By adjusting the installation position of the gripper base 402 on the slide 403 of the positive and negative bidirectional linear slide module 401, the limit distance of the mechanical gripper mechanism 406 in the X direction can be adjusted. By adjusting the position of the mechanical gripper mechanism 406 in the slide groove of the gripper support frame 404, the height of the mechanical gripper mechanism 406 in the Z direction can be adjusted. In conjunction with the sliding of the slide 403 of the positive and negative bidirectional linear slide module 401, packages of different sizes can be fixed.

[0104] The positive and negative thread bidirectional slide module 401 is composed of a lead screw nut 408, a positive and negative thread ball screw 409, a linear guide 410, a baffle 411, an aluminum alloy profile 412, a servo motor 413, a bearing seat 414, a synchronous wheel 415, a motor cover 416, a photoelectric switch 417, a guide rail slider 418, etc., and is fixed inside the second layer of the platform frame 3 by threaded connection and cylindrical pin interference fit.

[0105] Figure 7 is a schematic diagram of a robot control box in the present invention;

[0106] like Figure 7 As shown, the robot control box 5 includes an upper box 503, a lower box 504, a display screen 501, a control panel 502, and an industrial computer is installed inside the robot control box. The control panel communicates with the industrial computer through GPIO detection.

[0107] Figure 7 Schematic diagram of the visual guidance robot motion process in the present invention

[0108] Figure 8 It is a schematic diagram of the image processing algorithm flow in the present invention;

[0109] Fig. 9 Schematic diagram of the image processing algorithm flow in the present invention

[0110] like Figures 7 to 9 As shown, the two industrial cameras 105 and 107 of the present invention are connected to the input end of the industrial computer of the robot control box 5 through a video capture card; the dual-arm robot pan / tilt 201 and the head camera 202 are also connected to the industrial computer of the robot control box 5, and the multi-channel video signals are displayed in real time on the display screen 501 of the industrial computer of the robot control box. The robot control box display screen 501 and the robot control box control panel 502 constitute a human-machine interaction system of the robot control box industrial computer.

[0111] The computer NVIDIA Jetson TX2 is equipped with an Ubuntu system with a Linux kernel, and the ROS open source robot operating system is also installed as an information transmission channel.

[0112] The control cabinet includes PLC, multiple servo drivers for driving servo motors, transformers, air switches, contactors, relays, power filters, wiring terminals, etc. The electrical control board includes a switching power supply, NVIDIA jetson TX2 development board, video capture card, relays, power adapter, etc. The control box 5 includes an upper box 503, a lower box 504, a display screen 501, a control panel 502, and an industrial computer, etc. are installed inside the robot control box.

[0113] Among them, the NVIDIA jetson TX2 development board is responsible for data collection and processing of the realsense camera, robot arm motion posture solution and planning algorithm, control of dual-arm robot motion, control of the electric disassembly tool at the end of the rectangular coordinate robot, and as a host computer to communicate with the PLC motion controller as a slave computer. The slave PLC motion controller is responsible for controlling the servo driver to drive the servo motor to realize the movement of the linear slide module, receiving and processing photoelectric switch sensor data, and information communication with the host computer NVIDIA jetson TX2.

[0114] In order to ensure the safety of the bomb disposal personnel, the communication distance between the host computer and the slave computer is required to be far. The industrial computer of the robot control box is used as the host computer to communicate with the computer NVIDIA jetson TX2, and the computer NVIDIA jetson TX2 is used as the slave computer of the robot control box industrial computer. At the same time, the computer NVIDIA jetson TX2 is used as the host computer to control the dual-arm robot and its slave computer PLC motion controller to realize the control of the robot control box for the entire dismantling operation.

[0115] NVIDIA Jetson TX2 is mainly responsible for data collection and processing of RealSense D435, robot arm motion posture solution and motion path planning, information communication with the lower PLC motion controller, and communication with the upper industrial computer. RealSense D435 depth camera 116 is connected to NVIDIA Jetson TX2 via USB, NVIDIA Jetson TX2 is connected to the lower PLC motion controller via 485 bus and communicates via modbus protocol, the PLC motion controller in the electrical control cabinet 304 communicates with the servo driver via ethercat bus, and drives the linear slide module described in the present invention through the servo driver. Control the servo motor of the mechanical gripper mechanism 4, the rectangular coordinate robot system 1 and the walking axis linear slide module 207 of the dual-arm robot described in the present invention, and control the end disassembly tool of the rectangular coordinate robot system 1 through a relay. NVIDIA Jetson TX2 controls the servo motor of the dual-arm robot system 2 through CAN bus communication to control the movement of the dual-arm robot. The computer NVIDIA jetson TX2 communicates with the industrial computer in the upper robot control box 5 via ethernet, and all the above communications constitute the entire disassembly robot communication system.

[0116] A control method for an intelligent disassembly robot system comprises the following steps:

[0117] 1) The industrial computer controls the pan / tilt 201 and camera 202 of the dual-arm robot system 2, and displays multiple video signals on the display screen 501 in real time to monitor and observe the disassembly work, and sends control instructions to the electrical control board 308 through the industrial computer; the electrical control board 308 receives environmental information fed back by the depth camera 116;

[0118] 2) The depth camera 116 transmits the color image and the depth image of the environmental information of the target to be disassembled to the electrical control board, and the electrical control board processes the collected images to obtain the coordinates of the feature points of the target to be disassembled in the coordinate system of the robot working platform 3;

[0119] 3) The electrical control board 308 plans the motion of the mechanical arm of the dual-arm robot system 2 using a fast expansion random tree algorithm according to the coordinates of the feature points of the target to be disassembled in the coordinate system of the robot work platform 3, and obtains the planning path information;

[0120] 4) The electrical control board 308 obtains the rotation angles of each joint of the dual-arm robot system 2 through inverse kinematics solution according to the planned path information, and converts the rotation angles into control information of the motor;

[0121] 5) The electrical control board 308 controls the servo motor of the dual-arm robot through CAN bus communication to control the movement of the dual-arm robot and sends the control information of the motor to the PLC. The PLC controls the corresponding servo motor feed through the corresponding servo driver according to the control information of the motor to realize the control of the linear slide module.

[0122] The process of guiding the robot movement by the visual camera and the depth camera 116 is as follows: the realsense depth camera 116 is connected to the NVIDIA jetson Tx2 via USB, and the NVIDIA jetson Tx2 receives the color and depth information of the environment transmitted by the realsense depth camera. The left and right infrared sensors of the realsense d435 depth camera are mainly used for acquiring depth images and calculating depth information. The RGB sensor is used to acquire color images to correct the depth images. The external dot matrix transmitter actively emits infrared signals to increase the target brightness so that the depth data can be enhanced. The depth camera is fixed at the end of the rectangular coordinate robot 1. It is an Eye-in-Hand system. Using the official Intel SDK, point cloud images with RGB information can be obtained. After obtaining the initial point cloud image, the initial point cloud data needs to be processed to extract the disassembly target object from the disassembly operation plane and divide it into multiple independent point cloud sets. In the original point cloud image, the point cloud data is relatively complex as a whole, and there is a lot of noise. A large part of the point cloud data is invalid data. Using filtering, segmentation and other algorithms to process point cloud data can remove invalid data well. The straight-through filtering algorithm can remove invalid point cloud data such as the platform and objects on the platform. The voxel filtering algorithm can sparse the point cloud while ensuring the shape characteristics of the point cloud. The amount of point cloud data is reduced without significant changes in the point cloud structure, which significantly improves the subsequent point cloud processing speed. Then the SACSegmentation segmentation class can match and identify the platform in the point cloud according to the trained plane model and extract it. After that, the ConvexHull class and the ExtractPolygonalPrismData class extract the object point cloud on the platform according to the distribution of the platform point cloud. Finally, the EuclideanClusterExtraction clustering is used to segment the point clouds of multiple disassembled target objects into multiple independent small point cloud sets. After filtering, segmentation and other processing, the target object can be extracted from the environment. However, the feature point coordinates obtained at this time are based on the camera depth coordinate system. For the robot to operate the target object, it is necessary to obtain the coordinate value of the feature point in the coordinate system of the robot base. Therefore, it is necessary to perform a corresponding coordinate transformation on the feature point to obtain the coordinates of the feature point of the target object in the desktop coordinate system. After processing the captured image and obtaining the coordinates of the target object. In order to move the end effector of the robot from the initial position to the vicinity of the disassembly target, the computer NVIDIA jetson Tx2t is used to plan the path of the robot movement. The workspace-based rapid expansion random tree algorithm (RRT) is used for path planning to achieve the purpose of avoiding the target we want to avoid and moving to the target we want to move to. Assume C∈R 3is the Cartesian space of the robot arm in this paper. A point (x, y, z) in three-dimensional space is used to represent a coordinate in space. k is a random tree with k nodes, X init is the initial state, X goal is the target state. rand is a state point randomly selected in C space. Traverse T k , find the distance X rand The nearest node, distance X rand The nearest node is X nearst Then from X nearst and X rand Find the distance X in nearst X rand and X nearst The new node is closer to the node X. new , let p,q∈C, dist(p,q) denote the Euclidean distance between different positions. L is the search step length. If the search step length L≤dist(X nearest , X rand ), then in X nearst and X rand Find X on the connection line new , and dist(X nearest , X rand )=L. If there is X new If it is not in the obstacle area, a new node is added to the random tree. If it is in the obstacle area, a new set of X needs to be selected. rand As a candidate node, repeat the above process until the random tree T k There is a point in the equation that can eventually reach the target point. The extended formula is as follows:

[0123]

[0124] Due to X rand is randomly selected, so from X init To X goal The planned path is highly random and accidental. rand With X goal The direction vector of is highly correlated with the length of the path, so a vector representing the direction is added to the algorithm, making X rand Always approach the optimal path, the direction vector uses the current parent node as the reference node, X goal is the target node. Determine direct+point(X goal )-point(X nearest ) is the reference vector, 20 nodes are randomly selected, and X is calculated. rand With X goalThe angle between the direction of the connected straight line and the reference direction. Sort the angles by size and select the first four smallest ones as candidates. Use roulette to select the first four smallest random nodes to expand the subsequent path. This improved algorithm can shorten the planning time and improve the efficiency of the algorithm.

[0125] The algorithm flow is as follows:

[0126] Step 1: Set the starting point X init and the target point X goat , add environmental information such as obstacles.

[0127] Step 2: Expand the leaf nodes by expanding the formula.

[0128] a) Determine dist(X new ,X goal ). If L≥dist(X new ,X goal ) then the expansion ends and the loop exits.

[0129] b) If X is not achieved goal , expand the node, traverse T k , find X new . Select X rand Expand the leaf nodes. To reduce the chance, select 20 nodes as candidate random nodes X. rand , calculate dist(X rand ,X goal ), save to array

[20] , select the first four smallest random nodes X through roulette rand , and sort them from smallest to largest.

[0130] c) Enter 4 cycles, if dist(X new ,X rand )<L, then do not expand, reselect X rand If X is not found after 4 cycles, rand , then go to a). Determine X new Check whether the line connecting the node to its parent node touches any obstacle in the scene. If there is no collision, expand the node according to the aforementioned expansion formula and go to step a). When the loop ends, it is considered that the expanded leaf node is found and go to step a).

[0131] Step 3: Since the algorithm is performed in the workspace, the inverse kinematics of the robot arm is also required.

[0132] Step 4: After the inverse solution, the angles of all joints are obtained, and the forward solution is performed through the kinematic solution link, and finally the spatial coordinates corresponding to the joint angles are obtained, and collision detection is performed again.

[0133] Step 5: If there is a collision, randomly select a value between the angles of two adjacent points in the pose until a safe joint angle is obtained.

[0134] After obtaining the planned path information, the planned path information is used as the robot arm control information. Let the position of the robot end be x, here is a 6×1 vector, including the position information and posture information of the end of the robot arm. Assume that the joint angle of the robot arm is q=[q1,...,qi] T Then, the mapping relationship from the angle of the robot arm to the end position can be expressed as:

[0135] x=f(q)

[0136] Therefore, the inverse kinematics solution can be expressed as:

[0137] q=f -1 (x)

[0138] The numerical iteration method is selected to solve the inverse kinematics, and the Newton-Raphson iteration method is used to solve the inverse kinematics problem of the robot.

[0139] 1) Jacobian matrix of the robot

[0140] The Jacobian matrix of the robot arm is the mapping relationship between the robot arm's velocity in space and the joint velocity. Based on the Jacobian matrix, we can solve the end velocity of the robot arm from the joint velocity. The derivative of both sides of the formula x=f(q) with respect to time t is:

[0141]

[0142] In the formula is the Jacobian matrix of the robot arm, where is the velocity of the end of the robot in Cartesian space, is the joint velocity of the robot. After constructing the Jacobian matrix, the joint velocity can be inversely calculated based on the desired end velocity. The velocity of the robot end in Cartesian space and the joint velocity in joint space can be interchanged through the Jacobian matrix J(q). Since J(q) may be a matrix with non-full rank, the joint velocity is solved through the pseudo-inverse of Jacobian, that is:

[0143]

[0144] Among them, J # =J T (JJ T ) is the pseudo-inverse of the robot (J(q) is abbreviated to J).

[0145] 2) Differential motion

[0146] The differential motion formula means that when the amount of movement of the end of the robot arm is very small and the duration is very short, the slight change in the position and posture of the end of the robot arm is mathematically expressed by differential change. Assuming d is the differential translation and δ is the differential rotation, the differential motion of the end of the robot can be written as a six-dimensional column vector D = [d, δ] T =[dx,dy,dz,δx,δy,δz] T , then the differential motion can be expressed by the differential vector D as:

[0147] D=J(q)dq

[0148] Change the above formula to dq = J # D,

[0149] Assume that the current end position of the robot arm is T cur , when the end of the robot arm undergoes a slight change, according to the change relationship of forward and inverse kinematics, we can get:

[0150] T+dt=Trans(dx,dy,dz)Rot(k,dθ)T

[0151] Then dt = Trans(dx,dy,dz)Rot(k,dθ)TT = (Trans(dx,dy,dz)Rot(k,dθ))-I)T, where I is the unit matrix and the content in the brackets is the differential operator, so we can get Δ = Trans(dx,dy,dz)Rot(k,dθ)-I

[0152] 3) Newton-Raphson iteration method for inversion

[0153] Assume that the current end position of the robot arm is T cur , the desired target end position is T end , according to the above calculation method, it can be expressed as:

[0154] T end =T cur +ΔT cur =(I+Δ)T cur

[0155] in,

[0156] The main steps of using the Newton-Raphson iteration method are:

[0157] 1. Use forward kinematics and the current angle value qcur of the robot joint to calculate the current position and posture (position) Tcur of the end of the robot arm.

[0158] 2. Calculate the robot’s differential motion matrix Δ based on the robot’s target posture.

[0159] 3. Calculate the differential vector D of the terminal motion based on the current differential operator.

[0160] 4. Based on the current differential vector, calculate the Jacobian matrix J(q) of the robot arm.

[0161] 5. Find the size of dq, where the inverse of the Jacobian matrix is ​​the pseudo-inverse of the robot, as described above.

[0162] 6. According to the currently calculated dq, it is added to the current joint angle, that is, the joint angle changes to qcur=dq+qcur.

[0163] 7. Calculate the new terminal posture Tcur based on the new qcur, return to the first step, and continue iterating.

[0164] The above is the process of iterative calculation of the robot arm. At the same time, it is also necessary to design the termination condition for this iterative process. You can set the size of the maximum error of the actual accuracy, e, or the maximum number of iterations. When the two meet one of the conditions, the calculation is terminated. The mathematical expression is:

[0165] ||T cur -T end ||<eorn>N

[0166] Among them, n represents the current number of iterations, and N represents the set maximum number of iterations.

[0167] After obtaining the joint angles of the robot arm, the joint angles are controlled by the PID controller through the torque control of the motor, thereby realizing the control of the robot. In addition, NVIDIA Jetson Tx2 can also receive operator control information from the upper robot control box industrial computer interaction system, so that NVIDIA Jetson Tx2 can make decisions based on the feedback information of the 3D environment detection data and the operator control information.

[0168] The industrial computer of the robot control box 5 controls the pan / tilt 201 and camera 202 of the dual-arm robot 2, and displays multiple video signals in real time on the display screen 501 to monitor and observe the disassembly work and provide guidance for the entire disassembly operation. At the same time, the bomb disposal operator remotely uses the robot control box 5 to send control information to the NVIDIA jetson Tx2. The NVIDIA jetson Tx2 receives the feedback information of the depth camera and the control information of the robot control box 5. After completing the motion planning, it is transmitted to the drive system through bus communication. The drive system receives the control information and drives the servo motor to move. The mechanical arm motors and other motion units of the dual-arm robot 2 and the rectangular coordinate robot 1 receive different signals, and each motor moves to a different position, thereby controlling the dual-arm robot 2 and the rectangular coordinate robot 1 and each motion unit to achieve the expected motion target, and using the disassembly tool clamped by the robot to complete the entire disassembly operation.

[0169] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that deviate from the spirit and principles of the present invention shall be equivalent replacement methods and shall be included in the protection scope of the present invention.

Claims

1. An intelligent disassembly robot system, characterized in that: include: A robot working platform (3), and a rectangular coordinate robot system (1), a dual-arm robot system (2), a mechanical gripper mechanism (4), and a disassembly control system arranged on the robot working platform (3); A mechanical gripping mechanism (4) is provided between the rectangular coordinate robot system (1) and the dual-arm robot system (2); The disassembly control system is respectively connected to the rectangular coordinate robot system (1), the dual-arm robot system (2), and the mechanical gripper mechanism (4); The disassembly control system comprises an industrial computer, an electrical control panel (308), and a PLC connected in sequence; The PLC is connected to a plurality of servo drivers, and the plurality of servo drivers are respectively connected to servo motors of the rectangular coordinate robot system (1), the dual-arm robot system (2), and the mechanical gripper mechanism (4) to perform corresponding actions; The electrical control panel (308) is connected to the dual-arm robot system (2) via a servo driver to control the execution of the dual-arm robot system (2); the electrical control panel (308) is connected to the rectangular coordinate robot system (1) to receive environmental information of the target to be disassembled collected by the rectangular coordinate robot system (1); The electrical control board (308) is used to receive control instructions sent by the industrial computer to control the dual-arm robot system (2) to perform corresponding actions, and receive motor control information obtained by the electrical control board (308) after processing the control instructions and environmental information, and send it to the PLC to control the movement of the rectangular coordinate robot system (1) and the mechanical gripper mechanism (4); The control method of the intelligent disassembly robot system comprises the following steps: 1) The industrial computer controls the pan / tilt platform (201) and binocular camera (202) of the dual-arm robot (2), and displays multiple video signals on a display screen (501) in real time to monitor and observe the disassembly work, and sends control instructions to the electrical control board (308) through the industrial computer; the electrical control board (308) receives environmental information fed back by the depth camera (116) of the rectangular coordinate robot system (1); 2) The depth camera (116) transmits the color image and the depth image of the environmental information of the object to be disassembled to the electrical control board (308), and the electrical control board (308) processes the collected images to obtain the coordinates of the feature points of the object to be disassembled in the coordinate system of the robot working platform (3); 3) The electrical control board (308) plans the movement of the mechanical arm of the dual-arm robot system (2) according to the coordinates of the characteristic points of the object to be disassembled in the coordinate system of the robot working platform (3), and obtains the planning path information; 4) The electrical control board (308) obtains the rotation angle of each joint of the dual-arm robot system (2) through inverse kinematics solution according to the planned path information, and converts the rotation angle into control information of the motor; 5) The electrical control board (308) controls the servo motor of the dual-arm robot through CAN bus communication to control the movement of the dual-arm robot and sends the control information of the motor to the PLC. The PLC controls the corresponding servo motor feed through the corresponding servo driver according to the control information of the motor to realize the control of the linear slide module.

2. The intelligent disassembly robot system according to claim 1, characterized in that: The robot operation platform (3) is also provided with a platform pan / tilt platform (306) and a camera (307); There are two platform pan-tilt platforms (306), which are respectively arranged on one side vertically fixed on the robot working platform (3) and perpendicular to the movement direction of the dual-arm robot system (2); each of the platform pan-tilt platforms (306) is provided with a camera (307) capable of realizing a pitch function, and the camera (307) is connected to an industrial computer via a video acquisition card.

3. The intelligent disassembly robot system according to claim 1, characterized in that: The rectangular coordinate robot system (1) comprises: a disassembly tool box (103), an X-axis linear slide module (107), a cantilever Y-axis linear slide module (109), a Z-axis linear slide module (110), and a depth camera (116); The X-axis linear slide module (107) is installed on one side of the robot working platform (3); the slide seat A (108) of the X-axis linear slide module (107) is fixedly connected to the Z-axis linear slide module (110), so as to drive the Z-axis linear slide module (110) to perform linear motion along the X-axis direction through the slide seat A (108); The base of the cantilever Y-axis linear slide module (109) is mounted on the slide seat B (101) of the Z-axis linear slide module (110) via an adapter plate (111), so as to realize the linear motion of the cantilever Y-axis linear slide module (109) along the Z-axis via the slide seat B (101); The disassembly tool box (103) is fixed on the slide seat C (115) of the cantilever Y-axis linear slide module (109); Industrial cameras (105) are symmetrically arranged on both sides of the disassembly tool box (103), and the industrial cameras (105) are connected to the industrial computer via a video acquisition card, and are used to transmit multiple video signals to the industrial computer in real time, and display them on a display screen (501) connected to the industrial computer; The depth camera (116) is fixedly mounted on a card holder below the disassembly tool box (103) and connected to an electrical control panel (308) of a disassembly control system, and is used to collect environmental information of the object to be disassembled; The servo motors corresponding to the X-axis linear slide module (107), the cantilever Y-axis linear slide module (109), and the Z-axis linear slide module (110) are connected to the PLC.

4. The intelligent disassembly robot system according to claim 3, characterized in that: A conversion plate (114) is horizontally arranged on the slide seat A (108) of the X-axis linear slide module (107), and a vertical plate (112) is vertically arranged on the conversion plate (114). One side of the vertical plate (112) is fixedly connected to the conversion plate (114), and the other side of the vertical plate (112) is fixedly connected to the base of the Z-axis linear slide module (110), so that the Z-axis linear slide module (110) is perpendicular to the X-axis linear slide module (107) and moves linearly along the X-axis direction with the slide seat A (108).

5. The intelligent disassembly robot system according to claim 1, characterized in that: The dual-arm robot system (2) comprises: a pan-tilt platform (201), a binocular camera (202), a seven-degree-of-freedom robot arm (203), a two-degree-of-freedom trunk (204), a gripper (205), a raised seat (206), a walking axis linear slide module (207), and a slide seat D (208); The bottom end of the two-degree-of-freedom trunk (204) is fixed to the robot pad (206) by bolts, and the robot pad (206) is fixed to the slide D (208) of the walking axis linear slide module (207). The walking axis linear slide module (207) is arranged in the robot working platform (3), and the slide D (208) is arranged on the slideway on the surface of the robot working platform (3). The movement direction of the walking axis linear slide module (207) is parallel to the X-axis linear slide module (107), so that the slide D (208) drives the two-degree-of-freedom trunk (204) to move along the X-axis direction; A pan-tilt platform (201) is provided at the top of the two-degree-of-freedom trunk (204), and a binocular camera (202) is provided on the pan-tilt platform (201). The binocular camera (202) is connected to the input end of the industrial computer of the control box (5) through a video acquisition card, and is used to transmit multiple video signals to the control box (5) in real time, and display them on the display screen (501), so as to monitor and observe the disassembly work and provide guidance for the entire disassembly work; the output end of the control box (5) is connected to the pan-tilt platform (201), and is used to adjust the posture of the binocular camera (202) by controlling the pan-tilt platform (201); Seven-degree-of-freedom mechanical arms (203) are symmetrically arranged on both sides of the top of the two-degree-of-freedom trunk (204), and a clamping claw (205) is arranged at the end of the seven-degree-of-freedom mechanical arm (203); A mechanical gripping mechanism (4) is provided on the robot working platform (3) between the seven-degree-of-freedom mechanical arm (203) and the cantilever Y-axis linear slide module (109); The servo motor of the walking axis linear slide module (207) is connected to the corresponding servo driver in the control cabinet; the servo motors corresponding to the two-degree-of-freedom trunk (204) and the seven-degree-of-freedom mechanical arm (203) are respectively connected to the electrical control board (308) through the corresponding servo drivers.

6. The intelligent disassembly robot system according to claim 1, characterized in that: The mechanical gripper mechanism (4) comprises: a positive and negative thread bidirectional linear slide module (401), a gripper base (402), a slide seat E (403), a gripper support frame (404), a rubber part (405), and a mechanical gripper (406); The mechanical gripper (406) is slidably mounted on the gripper support frame (404), and there are two mechanical grippers (406) arranged opposite to each other. Rubber parts (405) are arranged on opposite sides of the mechanical gripper (406) to prevent the entire mechanical gripper from loosening when vibration and impact occur. There are two gripper support frames (404), which are respectively threadedly connected to the corresponding gripper bases (402); the gripper bases (402) are arranged on the corresponding positive and negative thread bidirectional linear slide module (401) through the slide seat E (403); The servo motor of the forward and reverse bidirectional linear slide module (401) is connected to the PLC via a corresponding servo driver.

7. The intelligent disassembly robot system according to claim 1, characterized in that: In step 2), the electrical control board (308) processes the collected images to obtain the coordinates of the object to be disassembled, including the following steps: After the depth camera (116) acquires the initial point cloud image; (2.1) Processing the initial point cloud data through straight-through filtering to remove invalid point cloud data; (2.2) The point cloud is sparsely made while ensuring the shape characteristics of the point cloud through the voxel filtering algorithm, so as to reduce the amount of point cloud data; (2.3) Point cloud segmentation: Match and identify the platform in the point cloud data according to the trained plane model; input the sparse point cloud image into the plane model, extract the plane point cloud data subset on the platform according to the distribution of the platform point cloud data, and obtain the plane point cloud map; The trained plane model is a model that recognizes a point cloud image of a plane from the entire environment image; (2.4) Point cloud clustering: Obtain all point clouds on the plane point cloud map, and divide the point cloud data of the target to be disassembled on the plane point cloud map into multiple independent small point cloud sets; obtain the coordinates of the feature points of the target to be disassembled based on the camera depth coordinate system; (2.5) Coordinate transformation: Perform coordinate transformation on the feature points of the target to be disassembled based on the camera depth coordinate system to obtain the coordinates of the feature points of the target object in the coordinate system of the robot work platform (3).

8. The intelligent disassembly robot system according to claim 1, characterized in that: The step 3) comprises the following steps: 3.1) Set the starting point X of the end of the robot arm init and the target point X at the end of the robot arm goal ,Add obstacle environment information; The obstacle environment information is: point cloud data of an object that blocks the end of the robot arm in the plane point cloud map to represent the obstacle; 3.2) Expand the leaf nodes by expanding the formula: a) Determine dist(X new ,X goal ), if L≥dist(X new ,X goal ) then the expansion is completed and this loop is exited; b) If L<dist(X new ,X goal ) means that the planned path does not reach the end point from the starting point, then an extension node is obtained and T is traversed. k , find X new ; Select X rand Expand the leaf nodes, that is, select 20 nodes as candidate random nodes X rand , calculate dist(X rand ,X goal ), save it to array[20], sort it from small to large, and select the first four smallest random nodes X through roulette wheel rand ; Among them, the T k is a random tree with k nodes representing the positions of the robot end, X rand is a state point randomly selected in the Cartesian space C of the robot; L is the search step length, and dist represents the Euclidean distance The search for X new , that is: find the distance X rand The nearest node X nearst ; then from X nearst and X rand Find the distance X in nearst X rand and X nearst The node X that is closer to new ; c) Enter 4 cycles, for each random node X rand , if dist(X new ,X rand )<L, then do not expand; if X is not found after completing 4 cycles rand , return to step a); Judge X new Check whether the line connecting the node to its parent node touches any obstacle in the scene. If no collision occurs, expand the node and return to step a). When the loop ends, it is considered that the expanded leaf node is found, and returns to step a); 3.3) Use the robot kinematics to find the inverse solution for each pose point of each path point to obtain the angles of all joints; 3.4) After the inverse solution, the angles of all joints are obtained, and the kinematic solution is performed to obtain the spatial coordinates corresponding to the joint angles, and collision detection is performed again; 3.5) If the inverse solution fails, randomly select a value between the angles of two adjacent points in the posture until a joint angle that satisfies the safety of the robot arm in this posture is obtained.

9. The intelligent disassembly robot system according to claim 1, characterized in that: The step 4) is specifically: According to the planned path information, the Newton-Simpson iteration method is used to perform kinematics solution to obtain the joint angles of the robot arm. The specific steps are as follows: 4.1) Using positive kinematics and the current angle value of a joint of the robot arm: q cur , get the position and posture of a joint at the end of the current robot arm, that is, the posture: T cur ; 4.2) According to the posture of a joint target of the robot arm, obtain the differential motion matrix Δ of a joint of the robot arm; 4.3) According to the differential operator in the current differential motion matrix Δ, the differential vector D of the robot end motion is obtained, and according to the current differential vector D, the differential motion of a joint of the robot arm is represented by the differential vector D; according to the inverse of the Jacobian matrix as the pseudo-inverse of the robot arm, the size of dq is calculated; 4.4) According to the currently calculated dq, it is added to the current joint angle, that is, the joint angle changes to q cur =dq+q cur ; 4.5) According to the q of the remaining joints of the robot arm cur Calculate the posture T of the remaining joint ends of the robot arm cur , return to 4.1) to continue iterating; Set the termination condition for the iteration process. When the actual maximum accuracy error e meets the threshold, or the maximum number of iterations n reaches the set number of iterations N, the iteration is terminated. The actual maximum accuracy error is: The current end position of the robot arm T cur The terminal pose T with the desired target end , the difference between them, namely: ||T cur -T end ||<e.

Images