A miniature chess robot AGV

Abstract

This invention discloses a miniature chess robot AGV, comprising a movement power system, a positioning system, and a decision control system communicating with a PC. The movement power system and the positioning system are controlled by the decision control system. The movement power system includes a motion device and a piece-carrying device. According to the instructions of the decision control system, the motion device drives the piece-carrying device to move on the chessboard, and the piece-carrying device performs the actions of moving or capturing pieces. The positioning system identifies and locates the movement power system, the chessboard, and the pieces on the chessboard, and transmits the positioning information to the decision control system. The decision control system plans the movement path, trajectory, and movement or capturing actions of the movement power system based on the movement information from the PC, and issues instructions to the movement power system. The miniature chess robot AGV of this invention has advantages such as compact size, flexible movement, high safety, and low hardware cost.

Description

Technical Field

[0001] This invention relates to the field of intelligent service equipment technology, and in particular to a miniature chess robot AGV. Background Technology

[0002] Intelligent chess robots are an emerging product in the intelligent gaming device market. The typical functional requirements for a chess robot are as follows:

[0003] (1) Using radio frequency technology, machine vision, etc., the chess pieces on the chessboard can be identified and accurately located, and the changes in the chessboard situation during the game can be automatically detected.

[0004] (2) The robot decision-making body receives the board information after the player makes a move and uses decision-making algorithms such as Alpha-Beta pruning game algorithm and Monte Carlo tree to determine the robot's next move.

[0005] (3) The decision-making body transmits the next move information to the move execution body through a certain communication mechanism. The move execution body converts the move information into a lower-level instruction, then moves to the designated piece to complete the move and capture action, and then moves to the designated position outside the chessboard to wait for the next instruction.

[0006] For chess-playing robots, most existing products on the market use robotic arms to complete the moves, such as the GRB Chinese Chess Robot from Golden Carrot Intelligent Technology and the Xianglian Intelligent Chess Robot System.

[0007] For example, Chinese patent document CN113926177A discloses a chess robot, including: a chessboard base with a chessboard on it; a chess piece recognition module for recognizing chess pieces placed on the chessboard; a drive mechanism including a first linear drive member disposed on the chessboard base along the X-axis, a second linear drive member disposed at the execution end of the first linear drive member along the Y-axis, and two third linear drive members disposed side by side at the drive end of the second linear drive member along the Z-axis; and two chess-grabbing claws disposed at the execution ends of the two third linear drive members for grabbing chess pieces. Chinese patent document CN110180164A discloses a chess robot, including a chessboard base, a robotic arm, and a vision component. The chessboard base includes a chessboard support, and the upper surface of the chessboard support is provided with a chessboard area and a chess piece storage area. The vision component includes a camera and a camera support, with the lower end of the camera support fixedly mounted on the chessboard support and the upper end of the camera support fixedly mounted on the camera. The robotic arm includes a forearm skeleton, a main arm skeleton, a tool end assembly, and a robotic arm support assembly. The robotic arm support assembly is fixedly mounted on the chessboard support, and the robotic arm support assembly is provided with a first power mechanism for driving the main arm skeleton to rotate. The main arm skeleton is provided with a second power mechanism for driving the forearm skeleton to rotate, and the free end of the forearm skeleton is fixedly mounted with the tool end assembly.

[0008] Existing chess robots generally use robotic arms as the chess-moving mechanism. Although well-manufactured and highly integrated robotic arms have superior performance in motion control, enabling the overall functions of the product to be realized well, the cost of the robotic arm will increase significantly with the increase of the control precision of the robotic arm. Chess robot manufacturers on the market are all facing the balance between the control precision of the robotic arm and the product cost, and the final price is in the tens of thousands of yuan. The price of the GRB Chinese Chess Robot is even as high as 49,800 yuan.

[0009] Although an intelligent chess robot incorporates many advanced technologies and can provide a good chess playing experience, the use of a robotic arm determines its high price, which is unaffordable for most individuals and families, thus hindering the promotion and development of this product. Summary of the Invention

[0010] This invention provides a miniature chess robot AGV, which has the advantages of small size, flexible movement, high safety and low hardware cost.

[0011] The technical solution of the present invention is as follows:

[0012] A miniature chess robot AGV includes a chess-moving power system, a positioning system, and a decision control system that communicates with a PC. The chess-moving power system and the positioning system are controlled by the decision control system.

[0013] The aforementioned chess-playing power system includes a motion device and a chess-moving device. According to the instructions of the decision control system, the motion device drives the chess-moving device to move on the chessboard, and the chess-moving device completes the action of moving or capturing pieces.

[0014] The positioning system identifies and locates the chessboard, the chessboard and the pieces on the chessboard, and transmits the positioning information to the decision control system.

[0015] The decision control system plans the movement path, trajectory, and move or capture actions of the movement power system based on the move information on the PC, and issues instructions to the movement power system.

[0016] The aforementioned motion device includes a vehicle body, wheels, and a motor fixed to the vehicle body and driving the wheels;

[0017] The chess-moving device includes:

[0018] The lifting mechanism is fixed to the vehicle body and drives the suction and placement mechanism to lift and lower according to instructions;

[0019] The pick-up and placement mechanism is fixed at the bottom of the lifting mechanism and picks up or places chess pieces according to instructions.

[0020] The chess-moving device works in conjunction with the motion device to complete the actions of moving or capturing pieces.

[0021] Preferably, the motor drives the wheels via gear transmission.

[0022] Preferably, the lifting mechanism includes:

[0023] The sliding rod is vertically fixed to the vehicle body;

[0024] A slider, which can move vertically, is mounted on a sliding rod;

[0025] The bracket is fixed to the vehicle body;

[0026] The crank is rotatably fixed to the bracket, with one end fixed to the slider and the other end connected to the servo motor.

[0027] A servo motor controls the height of one end of the crank, thereby controlling the movement of the slider along the slide bar.

[0028] Preferably, the suction and placement mechanism includes a suction cup and an air pump; the suction cup is fixed to the bottom of the lifting mechanism and connected to the air pump through an air guide tube; an electronic switch is provided on the air guide tube to control the operation of the suction cup according to the command.

[0029] The positioning system includes:

[0030] A fixed global camera above the chessboard acquires visual information about the movement power system, the chessboard, and the pieces.

[0031] An inertial measurement unit (IMU) fixed to the chess movement dynamic system is used to measure the motion information of the chess movement dynamic system;

[0032] RFID readers fixed to the chess movement power system and electronic tags placed on the chessboard;

[0033] The visual information acquired by the global camera and the motion information measured by the inertial measurement unit are subjected to extended Kalman filtering to obtain the current positioning information of the chess movement dynamic system; the RFID reader reads the electronic tag information to determine the position of the chess movement dynamic system, and fuses it with the visual information acquired by the global camera to correct the positioning of the chess movement dynamic system.

[0034] The positioning system described above locates the movement power system and chess pieces by including the following steps:

[0035] (a) The global camera acquires RGB images of the chess movement power system, chessboard and chess pieces and transmits them to the PC. The PC performs Canny edge detection, grayscale and binarization and Hough transform-based circular edge detection on the RGB images in sequence.

[0036] (b) Input the RGB image processed in step (a) into an image recognition model based on a convolutional neural network and trained to obtain the type of each piece and its position in the RGB image, the identification pattern of the movement dynamic system and its position in the RGB image;

[0037] (c) Using the position of the move dynamics system in the RGB image obtained in step (b), combined with the depth information and intrinsic parameter information of that point provided by the global camera, the exact position information (x, y) of the move dynamics system in the world coordinate system with the depth camera as the origin is calculated by the following formula:

[0038]

[0039] In this context, the pixel position of the chess movement dynamic system in the RGB image is denoted as (u,v), z is the depth information of point (u,v), and the intrinsic parameters include focal lengths fx and fy and principal point horizontal and vertical offsets cx and cy.

[0040] The decision control system receives the next move information generated by the PC and plans the motion path, trajectory, and move or capture actions of the move power system based on the move information.

[0041] The strategy for planning the move or capture actions of the chess-playing dynamic system in the decision control system includes:

[0042] (A1) Determine if the next move involves capturing a piece;

[0043] (A2) If there is no capturing step but only moving pieces on the board, the moving power system first moves to the position of the piece to be moved, controls the moving device to descend and pick up the piece, then controls the moving device to rise to a height higher than one piece, maintains the raised state to move to the target position, puts down the piece, and the moving device returns to the initial height.

[0044] (A3) If there is a capture step, the movement power system first moves to the captured piece, moves the captured piece to a designated position outside the board, and then moves to the capturing piece to move it to the original position of the captured piece.

[0045] The strategy for the decision control system to plan the motion path of the chess-playing dynamic system includes:

[0046] (B1) The movement propulsion system only moves in a straight line on the chessboard, and turning movements need to be completed outside the chessboard;

[0047] (B2) The movement propulsion system moves over the piece, and the distance between the wheels of the movement propulsion system is greater than the diameter of the piece.

[0048] (B3) After completing the move or capture action, the chess movement power system will move outside the chessboard to wait for the next instruction;

[0049] (B4) After receiving a move command outside the chessboard, the move power system will determine the position of a target piece, and then move from the current position to the target piece position to complete the move or capture action.

[0050] (B5) After completing the move or capture action on the chessboard, the movement power system will move along the horizontal or vertical straight line of the chessboard to the outside of the chessboard. If there is no need to place the captured piece, it will stop in place and wait for subsequent instructions. If there is a need to place the captured piece, it will plan a smooth curve to move to the designated placement area of ​​the captured piece.

[0051] Further preferred strategy (B4) includes:

[0052] (B4-1) If the current position of the movement dynamic system and the position of the target piece are in the same row or column parallel to the grid lines of the chessboard, and its current posture is facing or away from the position of the target piece, then the movement dynamic system can move directly to the position of the target piece through linear motion.

[0053] (B4-2) If the current pose of the chess moving power system is not the pose described in (B4-1), first plan a path and the corresponding motion trajectory outside the chessboard that can reach the pose described in (B4-1), and adjust the pose of the chess moving power system according to the plan. Then, move to the target chess piece position through linear motion.

[0054] The mathematical expression of strategy (B4) is:

[0055] Establish a plane coordinate system. In the chess moving plane where one player is above and the other is below, the positive direction of the x-axis is the far right of the chess moving plane, and the positive direction of the y-axis is the far bottom of the chess moving plane;

[0056] Record the current position of the chess moving power system as (xc, yc), the current pose as (xz, yz), and the target position as (xg, yg); It is set that the chess moving power system only moves in a straight line along the x-axis on the chessboard, so the target pose is recorded as (x - , 0) or (x + , 0);

[0057] Record the set of poses that satisfy the description in (B4-1) as the transition poses {(xi, yi), (xzi, yzi) | yi = yg, (xzi, yzi) ∈ (x - , 0) ∪ (x + , 0)}; Record the set of transition poses moving in the positive x-axis direction as the left transition area {(xi, yi), (xzi, yzi) | yi = yg, (xzi, yzi) ∈ (x + , 0)}; The set of transition poses moving in the negative x-axis direction is the right transition area {(xi, yi), (xzi, yzi) | yi = yg, (yzi, yzi) ∈ (x - , 0)};

[0058] If the current pose yc = yg and yz = 0, then when xc > xg, the chess moving power system is in the right transition area, and the chess moving power system reaches the target pose along the negative x-axis direction; when xc < xg, the chess moving power system is in the left transition area, and the chess moving power system reaches the target pose along the positive x-axis direction;

[0059] If the current pose yc ≠ yg or yz ≠ 0, then when xc > xg, the chess moving power system first moves along a planned smooth curve to the right transition area, and the transition poses are (xi, yg), (x - , 0), and then the chess moving power system reaches the target pose along the negative x-axis direction; when xc < xg, the chess moving power system first moves along a planned smooth curve to the left transition area, and the transition poses are (xi, yg), (x + , 0), and then the chess moving power system reaches the target pose along the positive x-axis direction.

[0060] Since the size and weight of the components selected in this invention are subject to significant limitations, and considering current manufacturing processes, the actual motion state of the chess-moving power system (including position, speed, acceleration, etc.) will deviate to some extent from the expected input of the actual input control signal. Preferably, during the movement of the chess-moving power system, the decision control system uses feedback control to plan the motion trajectory of the chess-moving power system.

[0061] The method of planning motion trajectory using feedback control includes:

[0062] The current pose, current velocity, and target position of the chess movement dynamics system are obtained from the positioning system.

[0063] Let the current pose (x-coordinate, y-coordinate, and attitude deflection angle) of the chess movement dynamic system be denoted as start=(x(t),y(t),θ(t)) T The target pose is goal = (xg, yg, θg). T The translational velocity is v(t), and the rotational velocity is w(t);

[0064] If we take the target position as the origin of the coordinate system and the target direction as the x-axis, then the motion model of the chess movement dynamic system in this coordinate system is:

[0065]

[0066] in Let x(t), y(t), and θ(t) represent the first derivatives of x(t), y(t), and θ(t) with respect to time, respectively. Hereinafter, x(t), y(t), θ(t), v(t), and w(t) will be abbreviated as x, y, θ, v, w.

[0067] Based on the idea of ​​reducing the distance and angle difference between the target pose and the target position, the Cartesian coordinate representation is converted to polar coordinate representation:

[0068] (ρ,α,β)T;

[0069] Where ρ is the distance between the moving dynamic system and the origin of the coordinate system, β is the angle between the direction of the line connecting the moving dynamic system and the origin of the coordinate system, and α is the angle between the direction of the moving dynamic system and the line connecting them, that is:

[0070]

[0071] β=arctan2(yg-y(t),xg-x(t)),

[0072] α = β - θ;

[0073] Thus, we obtain the rate of change of the polar coordinates ρ, α, β at this point. That is, the first derivative of the polar coordinates ρ, α, β with respect to time:

[0074]

[0075]

[0076] For system (1), the virtual control law is designed as follows:

[0077] α = arctan(-k1β),

[0078] Where k1 is a coefficient freely chosen based on the actual system, which allows the system (1) to move from any point to the origin, and further, a constant k is chosen. ρ And order:

[0079] v = k ρ ρ,

[0080] It can eliminate singularities, enabling the system to achieve asymptotic stability.

[0081] Let the difference z ≡ α - arctan(-k1β) (≡ means that z is always equal to the expression on the right-hand side of the equation), then the first derivative of the difference z with respect to time is expressed as:

[0082]

[0083] The formula Substituting, we get:

[0084]

[0085] Choose a constant k2, and define Then we have:

[0086]

[0087] When k2 > 1, the larger k2 is, the faster z approaches 0 (exponential change);

[0088] When k2 >> 1, the desired globally exponentially stable system is obtained.

[0089] Let path curvature

[0090]

[0091] Since rotational speed and translational speed are related, while path curvature is independent of translational speed, the translational speed can be adjusted based on the path curvature. Therefore, the following speed selection method exists:

[0092]

[0093] The values ​​of μ and λ need to be selected according to the actual situation.

[0094] Analysis of the above formula shows that:

[0095] When the current path is a straight line, k(ρ,α,β)→0, and at this time v=v max ;

[0096] When k(ρ,α,β)→∞, v→∞.

[0097] Therefore, after determining the appropriate values ​​of μ and λ, it is only necessary to adjust the values ​​of parameters k1 and k2 according to the motion constraints of the actual object.

[0098] By selecting the four parameters μ, λ, k1, and k2, a feedback control law is determined, as shown in the following equation:

[0099]

[0100]

[0101] w=k(ρ,α,β)v (3);

[0102] The motion control commands v and w given to the current chess movement dynamics system can be determined, and the rate of change for obtaining the current position information can be further calculated. Ultimately, the coordinates ρ, α, and β, which describe the position of the dynamic system of the move, will all become 0.

[0103] Since the dynamic system for chess moves typically does not need to move to the target position with perfect precision, but only to achieve a certain level of positional accuracy, this avoids excessive planning and saves planning time. Therefore, it is necessary to set an appropriate error limit for the numerical solution. Preferably, the relative error limit is 2%.

[0104] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0105] (1) The present invention adopts a gear transmission system, wireless communication and overall layered structural design, integrating a variety of micro-small mechanisms. While realizing the basic functions of AGV such as information reception, precise positioning and motion control, and object picking and placing, it ensures that it is small in size and highly flexible. It is suitable for chess robot to play chess and can be extended to special scenarios where obstacles are dense and space is limited in size.

[0106] (2) The present invention uses a miniature air pump, a suction cup and a connecting tube as the mechanism for picking up and placing objects. The mechanism is small in size, can pick up large objects, and the suction cup has a certain degree of flexibility, so the requirements for the shape of the object and the smoothness of the surface are low.

[0107] (3) Compared with the robotic arm grasping method, the present invention has a larger range of motion, lower hardware cost, and greater flexibility than a large-size AGV, thus having better scene adaptability.

[0108] (4) When this invention is used as the actuator of a service robot, its compact appearance and flexible movement mode have good visual appeal, and it is lighter and safer than a robotic arm, and has a better companion effect.

[0109] (5) The present invention has a reasonable structural and functional design and is supported by advanced and reliable technology. Attached Figure Description

[0110] Figure 1 This is a schematic diagram of the structure of a miniature chess robot AGV;

[0111] Figure 2 The diagram shows the structure of the chess movement power system, with (a) being the front view and (b) the back view.

[0112] Figure 3 A schematic diagram of the movement trajectory of a miniature chess robot AGV on a chessboard;

[0113] Figure 4 The diagram shows the path planning strategy of the miniature chess robot AGV on the chessboard. (a) shows the miniature chess robot AGV on the right side of the chessboard, and (b) shows the miniature chess robot AGV on the left side of the chessboard.

[0114] Figure 5 A path diagram for implementing feedback control in a simulated miniature chess robot AGV. Detailed Implementation

[0115] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be noted that the embodiments described below are intended to facilitate the understanding of the present invention and do not limit it in any way.

[0116] This invention provides a miniature chess robot AGV and its chess-playing control method. The miniature AGV with a self-designed structure completes the chess-playing action and can move stably on a chessboard with each square measuring 7cm*7cm. It also works well with the vision system and chess-playing decision system of the intelligent chess robot.

[0117] like Figure 1 As shown, the miniature chess robot AGV of the present invention includes a chess-playing power system, a positioning system, and a chess-playing decision control system. The chess-playing power system and the positioning system are both connected to and subject to the unified scheduling of the chess-playing decision control system. The miniature chess robot AGV integrates the micro-sized hardware mechanisms of the above systems in a layered manner. Its main frame structure includes wheels, legs, multi-layered roof panels, and support pillars at the four corners of the roof panels. The remaining hardware components are fixed on the multi-layered roof panels.

[0118] The movement system drives the vehicle to move, completing the actions of picking up, lifting, and placing the pieces. For example... Figure 2 As shown, the chess-playing power system has four stainless steel legs 1, which are fixed below the acrylic roof plate 2. Each leg 1 is secured to the roof plate 2 with two pairs of bolts and nuts to increase stability. The bottom of each leg 1 has an opening for an internal bearing. Wheels 3 and the second transmission gear 4 are rotatably mounted on the leg 1 using rubber washers, metal shims, and metal shafts. The rubber washers and metal shims allow for a certain distance between the wheel 3 and the bearing without using large bearing seats or complex processes, ensuring flexibility while limiting the horizontal movement of the bearing to prevent it from slipping off the leg 1, thus improving motion performance. Due to current design limitations, four DC motors cannot be used for driving; therefore, only two legs not diagonally opposite each other require the installation of the second transmission gear 4 to achieve forward, backward, and turning movements of the car through differential operation. The DC motor 5 is fixed on the vehicle floor 2, which drives the first transmission gear 6 to rotate. The first transmission gear 6 meshes with the second transmission gear 4, thereby driving the second transmission gear 4 and the wheel 3 fixed to the second transmission gear 4 to rotate.

[0119] A suction cup 8 is vertically fixed to the bottom of the crank slider 7. It moves vertically via a copper column 9 fixed to the vehicle floor 2. The crank slider 7 and suction cup 8 are raised and lowered by a servo motor 10, which rotates to control the angle of rotation. The servo motor 10 is fixed to the vehicle center plate 11, which is in turn fixed to the vehicle floor 2 by the copper column. Alternatively, a stepper motor and a lead screw with a vertically fixed suction cup can be used to raise and lower the suction cup.

[0120] In this embodiment, the main controller 12 is a Raspberry Pi 3B+, which communicates wirelessly with a general-purpose PC via a Wi-Fi or Bluetooth module. It receives move commands (e.g., moving a cannon to capture a horse) from the PC and converts these commands into control information for the DC motor and lifting device. This coordinates with other modules to precisely move the pieces to the target position and perform the actions of picking up and placing them. The main controller 12 controls the rotation of the servo motor 10 and the on / off state of the air pump 13 and electronic switch via PWM waves, and drives the DC motor through the motor drive module. The motor drive module and the breadboard 14 are fixed to the middle plate 11 of the vehicle.

[0121] The roof panel 15 is fixed to the middle panel 11 of the vehicle by copper pillars, the main controller 12 is fixed to the lower surface of the roof panel 15, and the power supply battery 16 and air pump 13 are fixed to the roof panel 15.

[0122] The DC motor 5, the first transmission gear 6, the second transmission gear 4, the bearing, and the connecting shaft form a motion mechanism; the copper column 9, the crank slider 7, and the servo motor 10 form a lifting mechanism; the air pump 13, the suction cup 8, the air guide pipe connecting the air pump 13 and the suction cup 8, and the electronic switch set on the air guide pipe form a suction and placement mechanism.

[0123] The AGV's move decision control system includes a main controller, a motor drive module, a communication module, and other electrical connection components. Considering the AGV's size, weight, and performance requirements, an embedded microprocessor (MCU) is selected as the main controller. The main controller receives the robot's next move generated by the PC-based decision system and plans the movement path, trajectory, and move / pick up moves based on this information.

[0124] The strategies for planning moves in a chess decision control system include:

[0125] 1. First, determine whether the next move involves capturing a piece;

[0126] 2. If there is no capturing step and only the pieces on the board are moved, the car first moves to the position of the piece to be moved, controls the servo angle to lower the crank slider, lowers the suction cup, turns on the air pump to pick up the piece, then controls the servo to raise the suction cup and the piece together to a height higher than one piece through the crank slider, maintains the raised state to move to the target position, lowers the suction cup and the piece, the air pump stops working, the piece falls on the board, and the suction cup is raised back to its initial height;

[0127] 3. If there is a capture step, the trolley first moves to the captured piece, then moves the captured piece to a designated position outside the board, and then moves to another piece to move it to the original position of the captured piece.

[0128] The strategies for planning movement paths in a chess decision control system include:

[0129] 1. Because the chess pieces are densely distributed on the chessboard, the AGV must complete its turning action outside the chessboard in order to avoid colliding with the pieces when turning. The AGV only moves in a straight line on the chessboard.

[0130] 2. When the AGV moves in a straight line on the chessboard, the positional relationship between the chess pieces and the AGV is as follows: Figure 3 As shown. The AGV moves over the chess piece to... Figure 3 The dotted line in the middle serves as the movement track, and the distance between the wheels can be passed through the chess pieces;

[0131] 3. To avoid the AGV cart obscuring the chessboard for an extended period and hindering chess piece recognition, the AGV cart will move outside the chessboard after moving a piece and wait for the next movement instruction;

[0132] 4. After receiving a move command outside the chessboard, the AGV will determine a target movement position and then move from its current position to the target piece's location. The specific path planning strategy is as follows:

[0133] 1) If the current position of the AGV is in the same row or column as the target position, which is parallel to the chessboard grid, and its current posture is facing or away from the target position, then the AGV can move directly to the target position by linear motion.

[0134] 2) In most cases, the current position of the AGV is not in the same row or column as the target position. Therefore, firstly, plan a path and corresponding movement trajectory outside the chessboard that can directly reach the pose described in 1), move to the corresponding pose, and then move to the target position by linear movement.

[0135] 3) The mathematical expression of the above path planning strategy is:

[0136] First, establish a planar coordinate system. Define a movement plane where the red player is below the paper and the black player is above the paper. The positive x-axis points to the right of the paper, and the positive y-axis points to the bottom of the paper. For example... Figure 4 As shown.

[0137] remember:

[0138] Current location is

[0139] (xc,yc),

[0140] Current posture is

[0141] (xz,yz),

[0142] Target location is

[0143] (xg,yg),

[0144] Assuming the AGV can only move in a straight line along the x-axis on the chessboard, the target posture is denoted as...

[0145] (x - ,0) or (x + ,0),

[0146] The set of poses that satisfy 1) and are facing or away from the target position is denoted as the transition pose.

[0147] {(xi,yi),(yzi,yzi)|yi=yg,(xzi,yzi)∈(x - ,0)∪(x + ,0)},

[0148] remember:

[0149] The set of transition poses moving along the positive x-axis is called the left transition region.

[0150] {(xi,yi),(xzi,yzi)|yi=yg,(xzi,yzi)∈(x + ,0)},

[0151] The set of transition poses that move along the negative x-axis is called the right transition region.

[0152] {(xi,yi),(xzi,yzi)|yi=yg,(xzi,yzi)∈(x - ,0)}.

[0153] If the current pose

[0154] yc = yg and yz = 0,

[0155] Then when

[0156] xc>xg,

[0157] The AGV is located in the right transition zone, and then the AGV reaches the target pose along the negative x-axis.

[0158] when

[0159] xc <xg,

[0160] The AGV is located in the left transition zone and reaches the target pose along the positive x-axis.

[0161] If the current pose

[0162] yc≠yg or yz≠0,

[0163] Then when

[0164] xc>xg,

[0165] The AGV first moves along a planned smooth curve to the right transition zone, and the transition posture is...

[0166] (xi,yg),(x - ,0),

[0167] The AGV then reaches the target pose along the negative x-axis, such as... Figure 4 As shown in (a);

[0168] when

[0169] xc <xg,

[0170] The AGV first moves along a planned smooth curve to the left transition zone, and the transition posture is...

[0171] (xi,yg),(x + ,0),

[0172] The AGV then reaches the target pose along the positive x-axis, such as... Figure 4 As shown in (b).

[0173] 5. After the trolley has picked up or placed a piece on the chessboard, it will first move in a straight line along the x-direction to the outside of the chessboard. If there is no need to place a piece to be captured, it will stop in place and wait for subsequent instructions. If there is a need to place a piece to be captured, it will plan a smooth curve to move to the designated placement area of ​​the piece to be captured.

[0174] The positioning and decision-making control system of the miniature chess robot AGV works in tandem to achieve high chess move accuracy, including:

[0175] 1. The positioning system achieves millimeter-level positioning accuracy, and the current positioning is obtained by combining global camera visual information and motion information provided by IMU through extended Kalman filtering;

[0176] 2. An RFID reader is installed on the AGV body to continuously read data at a certain frequency. Electronic tags are placed at certain selected locations within the movement range. When the reader on the AGV continuously detects the electronic tag within a certain period of time, it indicates that the AGV is passing through the location of the electronic tag. The location of the AGV can be determined based on the read tag information. At this time, this location information can be used to replace the IMU information and be fused with the camera visual information to correct the positioning of the AGV.

[0177] 3. The decision control system uses real-time feedback control based on positioning information as a strategy for planning motion trajectories, resulting in a second-order smooth motion trajectory, thus compensating for the adverse effects of hardware process limitations on motion control accuracy.

[0178] The positioning system includes a depth camera located above the chessboard and fixed marker patterns on the exterior of the car. The camera captures images of the chessboard and transmits them to a PC. Using machine vision algorithms such as image binarization and Hough detection, the marker patterns on the car are identified and accurately located. The positioning accuracy should reach the millimeter level. The position information is transmitted to the control system to provide the current pose (xc, yc) and (xz, yz) for the feedback control of the car's movement.

[0179] The positioning system is based on the vision system of the chess robot, including a PC-based machine vision algorithm, a depth camera, a camera mounting bracket, and other electrical connections. It identifies and distinguishes the chess pieces and AGV markers on the chessboard, and uses this visual information and the camera model to determine the precise location of the AGV. The specific implementation steps are as follows:

[0180] Step 1 (Image Processing): On the PC, perform Canny edge detection, grayscale and binarization, and a circular edge detection algorithm based on Hough transform on the RGB image captured by the depth camera.

[0181] Step 2 (Image Recognition): Based on the image recognition model trained by the convolutional neural network, input the image processed in Step 1, and identify and distinguish the types of chess pieces and their positions in the image, as well as the AGV markers and their positions in the image;

[0182] Step 3 (Inverse Positioning): Using the pixel position of the AGV in the RGB image obtained in Steps 1 and 2, the coordinates of this pixel in the 2D image are (u, v). Combining the depth information z and intrinsic parameters provided by the depth camera, namely the focal lengths fx and fy and the horizontal and vertical offsets of the principal point cx and cy, the exact position of the AGV in the world coordinate system with the camera as the origin is calculated by the following formula.

[0183]

[0184] The strategy for the decision control system to plan motion trajectories is as follows:

[0185] Due to the significant limitations in the size and weight of the components selected for this invention, and considering current manufacturing processes, the actual motion state of the AGV (including position, speed, acceleration, etc.) will deviate to some extent from the expected input of the actual input control signal.

[0186] Therefore, the motion process employs feedback control to plan the trajectory. Specifically:

[0187] The AGV's current pose, current speed, and target position are calculated using machine vision. Let the AGV's current pose (horizontal coordinate, vertical coordinate, and attitude deflection angle) be denoted as...

[0188] start=(x(t),y(t),θ(t)) T ,

[0189] The target pose (x-axis, y-axis, attitude deflection angle) is:

[0190] goal=(xg,yg,θg) T ,

[0191] Translation speed is

[0192] v(t),

[0193] Rotation speed is

[0194] w(t),

[0195] If we take the target point as the origin of the coordinate system and the target direction as the x-axis, then the motion model of the AGV in this coordinate system is:

[0196]

[0197] in Let x(t), y(t), and θ(t) represent the first derivatives of x(t), y(t), and θ(t) with respect to time, respectively. Hereinafter, x(t), y(t), θ(t), v(t), and w(t) will be abbreviated as x, y, θ, v, w.

[0198] Based on the idea of ​​reducing the distance and angle difference between the target pose and the Cartesian coordinate representation, the representation is converted to polar coordinate representation.

[0199] (ρ,α,β) T

[0200] Where ρ is the distance between the AGV and the origin of the coordinate system, β is the angle between the direction of the line connecting the AGV and the origin of the coordinate system, and α is the angle between the direction of the AGV and the connecting line.

[0201]

[0202] β=arctan2(yg-y(t),xg-x(t)),α=β-θ

[0203] Thus, we obtain the rate of change of the polar coordinates ρ, α, β at this point. That is, the first derivative of the polar coordinates ρ, α, β with respect to time:

[0204]

[0205]

[0206] For system (1), the virtual control law is designed as follows:

[0207] α = arctan(-k1β)

[0208] Where k1 is a coefficient freely selected based on the actual system.

[0209] This allows the system (1) to move from any point to the origin, and further, a constant k can be chosen. ρ and order

[0210] v = k ρ ρ

[0211] It can eliminate singularities, enabling the system to achieve asymptotic stability.

[0212] Let the difference

[0213] z ≡ α - arctan(-k1β) (≡ indicates that z is always equal to the expression on the right-hand side of the equation).

[0214] Then there is

[0215]

[0216] The formula Substitute to get

[0217]

[0218] Choose a constant k2, and define

[0219]

[0220] Then there is

[0221]

[0222] When k2 > 1, the larger k2 is, the faster z approaches 0 (exponential change);

[0223] When k2 >> 1 (>> means much greater than), the desired globally exponentially stable system is obtained.

[0224] Let path curvature

[0225]

[0226] Since rotational speed and translational speed are related, while path curvature is independent of translational speed, the translational speed can be adjusted based on the curvature. Therefore, the following speed selection method exists:

[0227]

[0228] The values ​​of μ and θ need to be selected according to the actual situation.

[0229] Analysis of the above formula shows that:

[0230] When the current path is a straight line, k(ρ,α,β)→0, and at this time v=v max ;

[0231] When k(ρ,α,β)→∞, v→∞.

[0232] Therefore, after determining appropriate values ​​for μ and λ, it is only necessary to adjust the values ​​of parameters k1 and k2 according to the actual motion constraints of the object. The specific parameter values ​​selected in this implementation scheme are as follows, and there is a large adjustment range, which can be flexibly adjusted according to the actual situation:

[0233] parameter Value μ 0.5 λ 1.2 <![CDATA[k1]]> 1.2 <![CDATA[k2]]> 4.5 .

[0234] After selecting the four parameters μ, λ, k1, and k2, a feedback control law is determined, which is based on the following formula:

[0235]

[0236]

[0237] w=k(ρ,α,β)v (3)

[0238] The current motion control commands v and w given to the AGV can be determined, and the rate of change of the current position information can be obtained. Ultimately, the three coordinate values ​​ρ, α, and β describing the AGV's position all need to be reduced to 0. Since the AGV typically does not need to move to the target position with perfect precision, but only to achieve a certain level of positional accuracy, this avoids excessive planning iterations and saves planning time. Therefore, a suitable error limit needs to be set for the numerical solution (a relative error limit of 2% is used in this implementation).

[0239] The image of simulated feedback control is as follows Figure 5 As shown.

[0240] The gear transmission design and precise feedback control algorithm of this invention enable the micro AGV to have good motion performance during movement and raise the height of the DC motor. Simultaneously, the main controller and other electrical connections are layered and placed on the upper layer of the AGV to avoid damage to the chessboard due to violent collisions with the pieces. During movement, only the wheels of the micro AGV contact the chessboard; the height of all other components is above the pieces. Compared to using a robotic arm, using a micro AGV as the execution mechanism for chess robot movement significantly reduces costs and offers better safety and viewing experience.

[0241] This invention works in conjunction with a chess robot vision system and a decision-making system. The vision system identifies the current chessboard information, performs type identification and precise positioning of the pieces on the board. One hardware implementation of the vision system is a depth camera system located above the chessboard. If this camera system is used, it can be used simultaneously for the positioning of the AGV and the imaging of the chessboard. The decision-making system generates the robot's next move based on the chessboard information, converts the decision into move information, and transmits it to the control system of the micro AGV. Based on the move information, the control system provides feedback control to the movement of the AGV by controlling the speed of the DC motor until it reaches the target point for picking up or placing pieces. Then, it controls the air pump, electronic switch, stepper motor, or servo motor to collaboratively pick up, lift or place, and stop picking up the pieces, ultimately completing the move.

[0242] The embodiments described above provide a detailed explanation of the technical solutions and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A miniature chess robot AGV, characterized in that, The chess moving power system, the positioning system and the decision control system communicating with the PC are included, and the chess moving power system and the positioning system are controlled by the decision control system; The chess moving power system includes a moving device and a chess moving device, the moving device drives the chess moving device to move on the chessboard according to the instruction of the decision control system, and the chess moving device completes the chess moving or eating action; the moving device includes a vehicle body, wheels and a motor fixed on the vehicle body and driving the wheels; the chess moving device includes a lifting mechanism fixed on the vehicle body and driving the suction and placement mechanism to lift according to the instruction; the suction and placement mechanism is fixed on the bottom end of the lifting mechanism and sucks or places the chess pieces according to the instruction; the chess moving device and the moving device cooperate to complete the chess moving or eating action; The positioning system identifies and locates the chess moving power system, the chessboard and the chess pieces on the chessboard, and transmits the location information to the decision control system; The decision control system plans the moving path, trajectory and chess moving or eating action of the chess moving power system according to the chess moving information of the PC, and sends the instruction to the chess moving power system; The strategy of the decision control system planning the moving path of the chess moving power system includes: (B1) the chess moving power system only makes straight line motion on the chessboard, and the turning action needs to be completed outside the chessboard; (B2) the chess moving power system moves above the chess pieces, and the wheel spacing of the chess moving power system is greater than the diameter of the chess pieces; (B3) the chess moving power system moves to outside the chessboard to wait for the next instruction after completing the chess moving or eating action; (B4) the chess moving power system determines a target chess position after receiving the chess moving instruction outside the chessboard, and then moves from the current position to the target chess position to complete the chess moving or eating action; (B5) the chess moving power system moves to outside the chessboard along the horizontal or vertical straight line of the chessboard after completing the chess moving or eating action, stops at the original position and waits for the subsequent instruction if the eaten chess pieces need not be placed, or plans a smooth curve to the designated placement area of the eaten chess pieces.

2. The microchess robot AGV according to claim 1, characterized in that, The lifting mechanism includes: a slide rod vertically fixed on the vehicle body; a slide block vertically movable mounted on the slide rod; a bracket fixed on the vehicle body; a crank rotatable fixed on the bracket, one end of which is fixed with the slide block and the other end of which is connected with the rudder; a rudder controlling the height of one end of the crank to control the movement of the slide block along the slide rod.

3. The microchess robot AGV according to claim 1, wherein, The suction and placement mechanism includes a suction cup and an air pump; the suction cup is fixed on the bottom end of the lifting mechanism and communicates with the air pump through the air guide pipe; an electronic switch is arranged on the air guide pipe to control the working switch of the suction cup according to the instruction.

4. The microchess robot AGV according to claim 1, wherein, The positioning system includes: a global camera fixed above the chessboard to obtain the visual information of the chess moving power system, the chessboard and the chess pieces; an inertial measurement unit fixed on the chess moving power system to measure the motion information of the chess moving power system; an RFID reader fixed on the chess moving power system and an electronic tag arranged on the chessboard; The visual information obtained by the global camera and the motion information measured by the inertial measurement unit are subjected to extended Kalman filtering to obtain current positioning information of the chess moving power system; the RFID reader reads the electronic tag information to determine the position of the chess moving power system, and the visual information obtained by the global camera is fused to correct the positioning of the chess moving power system.

5. The microchess robot AGV according to claim 4, characterized in that, The positioning system for the chess moving power system and the chess pieces comprises the following steps: (a) the global camera obtains RGB images of the chess moving power system, the chessboard and the chess pieces and transmits them to the PC end, and the PC end sequentially performs Canny edge detection, grayscale and binaryzation, and circular edge detection processing based on Hough transformation on the RGB images; (b) the RGB images processed in step (a) are input into an image recognition model based on a convolutional neural network and trained to obtain the types of the chess pieces and their positions in the RGB images, the identification pattern of the chess moving power system and its position in the RGB images; (c) the position of the chess piece power system in the RGB image obtained by step (b) combined with the depth information of the position of the chess piece power system in the RGB image provided by the global camera and the intrinsic information, the exact position information of the chess piece power system in the world coordinate system with the depth camera as the origin is calculated by the following formula : ； Wherein, the pixel point position of the chess move power system in the RGB image is recorded as , The depth information of the point , the internal parameter information includes the focal length and the horizontal and vertical offset of the principal point .

6. The microchess robot AGV according to claim 1, wherein, The decision control system plans the strategy of the chess moving power system for the chess moving or eating action, which comprises: (A1) judging whether there is an eating step in the next chess moving step; (A2) if there is no eating step but only the movement of the chess pieces on the chessboard, the chess moving power system first moves to the position of the chess piece to be moved, controls the chess moving device to lower and pick up the chess piece, then controls the chess moving device to rise to a height higher than that of a chess piece, maintains the rising state to move to the target position, and lowers the chess piece, and the chess moving device returns to the initial height; (A3) if there is an eating step, the chess moving power system first moves to the eaten chess piece to move it to a designated position outside the chessboard, and then moves to the chess piece to be eaten to move it to the original position of the eaten chess piece.

7. The microchess robot AGV according to claim 1, wherein, Strategy (B4) comprises: (B4-1) if the current position of the chess moving power system and the target chess piece position are in the same row or column parallel to the grid lines of the chessboard, and the current posture of the chess moving power system is directly opposite or back to the target chess piece position, the chess moving power system can directly move to the target chess piece position through straight line motion; (B4-2) if the current position of the chess moving power system is not the position described in (B4-1), a path and the corresponding motion trajectory that can reach the position described in (B4-1) are planned outside the chessboard, and the position of the chess moving power system is adjusted according to the planning, and then the chess moving power system moves to the target chess piece position through straight line motion.

8. The microchess robot AGV according to claim 1, wherein, During the movement of the chess moving power system, the decision control system plans the motion trajectory of the chess moving power system by using feedback control method.

Images