A multi-robot kinematics self-calibration method and system based on visual measurement

Abstract

The application discloses a multi-robot kinematics self-calibration method and system based on visual measurement, and the method comprises the following steps: establishing a kinematics model of the whole pose adjustment mechanism; measuring the spatial coordinates of the wing end target point through a multi-view visual measurement system; selecting a certain POGO column, controlling the Z-axis movement of the POGO column, and making the X and Y axes do force following movement; fixing the axes of the other two POGO columns; recording the movement amount of each axis and the spatial coordinates of the wing end target point at m positions; and calculating the rotation axis of the wing, i.e. the connecting line of the centers of the ball heads of the two fixed POGO columns, through the SVD method. Selecting other POGO columns to move and repeating the above steps can obtain the equations of the connecting lines of the ball head centers of three groups of POGO columns. The intersection points of the three groups of straight lines can be obtained through solving, and the coordinates of the three ball heads can be obtained. Through the above multiple groups of movement data, an optimization model of the kinematics parameters is established, and the model is solved to obtain the calibrated kinematics parameters. The application has the characteristics of rapid calibration and automation, and is suitable for flexible assembly of the aircraft wing and fuselage.

Description

Technical Field

[0001] This invention relates to the field of intelligent assembly manufacturing calibration, and in particular to a multi-robot kinematics self-calibration method and system based on vision measurement. Background Technology

[0002] Wing-body docking is a critical step in aircraft final assembly. It typically employs multiple CNC positioners connected in parallel to form a 6-DOF (DoF) attitude adjustment mechanism, enabling six-DOF attitude adjustment between the aircraft and the wing. For wing adjustment equipment, a combination of three three-axis CNC positioners (POGO posts) is generally required to form the 6-DOF attitude adjustment mechanism. Typically, the three POGO posts are fixed to the ground, and their attitude parameters are obtained through calibration to establish the kinematic model of the entire adjustment mechanism. To improve assembly efficiency and production line flexibility, POGO posts are bound to AGVs to form mobile POGO posts, adaptable to wing-body docking of various product models, significantly enhancing the flexibility of the production line.

[0003] The mobile POGO column is positioned by an AGV, but its positioning accuracy is far lower than that required by the docking mechanism. Therefore, the initial pose of the POGO column can be considered unknown. The key to achieving wing pose adjustment lies in how to connect the POGO column's ball joint to the ball joint under the wing under these unknown conditions, and in obtaining the initial pose of the POGO column. Summary of the Invention

[0004] The purpose of this invention is to propose a multi-robot kinematic self-calibration method and system based on vision measurement, which realizes automatic connection between the equipment and the wing, and automatically establishes the kinematic parameters of the entire attitude adjustment mechanism to achieve precise motion control of the wing.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0006] A vision-based kinematic self-calibration method for multi-robot kinematics, applicable to a multi-robot system for wing-body docking of an aircraft, the multi-robot system comprising three mobile CNC positioners, a wing, a multi-view vision measurement system, visual target points at the wingtips and fuselage sides, a fuselage, and a support frame, including:

[0007] Step 1: Establish a kinematic model for the docking of multiple robot systems;

[0008] Step 2, based on the kinematic model, calibrate the docking parameters of the multi-robot system; including:

[0009] Step 2-1: Obtain a set of spatial coordinate data of the target point on the wing side through a multi-view vision measurement system, determine the homogeneous coordinate transformation matrix from the coordinate system of the i-th target point on the wing side to the coordinate system of the interface on the wing side, obtain the spatial coordinate values ​​of the target point on the wing side in the initial state, and the position of each motion axis of the mobile CNC positioner.

[0010] Step 2-2: Control the Z-axis movement of the i-th movable CNC positioner, while keeping the other two movable CNC positioners fixed. Record the position of each axis of the i-th movable CNC positioner and the spatial coordinates of the target point on the wing side at m positions. Control the movable CNC positioner to return to the initial position, replace the moving movable CNC positioner, and repeat this step until all movable CNC positioners have completed their execution.

[0011] Steps 2-3: Using the spatial coordinates of the target point on the wing side in the initial state and the spatial coordinates of the target point on the wing side at m positions, solve the homogeneous transformation matrix of the target point or the entire wing relative to the initial position using the SVD method.

[0012] Steps 2-4 involve solving for the approximate equation of the straight line using multiple sets of homogeneous transformation matrices.

[0013] Steps 2-5: Find the intersection points of the lines containing the three rotation axes;

[0014] Steps 2-6: Use the above data to solve for other parameters.

[0015] Furthermore, a kinematic model for multi-robot system docking is established, including:

[0016] In a multi-view vision measurement system, a camera coordinate system {C} is established, where the output target point's 3D coordinates are all values ​​in the camera coordinate system. A fuselage coordinate system {F} is established based on the geometric center of the fuselage-side interface, and a wing coordinate system {W} is established based on the geometric center of the wing-side interface. A coordinate system {V} is established at the center of the k-th target point on the wing side. k Establish a coordinate system {B} at the center of the i-th ball joint under the wing. i Establish a coordinate system {P} at the center of the ball socket of the i-th moving CNC positioner. i A coordinate system {G} is established at the center of the base of the i-th mobile CNC positioner. i}; where i=1,2,3; k=1...n, and n is the number of target points;

[0017] Let the camera coordinate system {C} in the multi-view vision measurement system be connected to the base coordinate system {G} of the i-th moving CNC positioner. i The homogeneous coordinate transformation matrix of} is ,in From coordinate system {C} to coordinate system {G} i The rotation transformation matrix of} Euler angles for rotational transformation, From coordinate system {C} to scale system {P} i Translation vector of}; base coordinate system {G} i} to the spherical-socket coordinate system {P i The homogeneous coordinate transformation matrix of} is ,in For coordinate system {G i} to the standard system {P i The translation vector of};

[0018] Let the transformation matrix from coordinate system {Pi} to coordinate system {Bi} be... The unknown is Euler angles. Coordinate system {B i} to coordinate system {W i The transformation matrix of} is ,in For coordinate system {B i} to coordinate system {W i The translation vector of};

[0019] Obtain coordinate system {C} to coordinate system {W} i The homogeneous transformation matrix of} is:

[0020] ;

[0021] The system of equations is obtained by rearranging the equations. ;

[0022] get ;

[0023] By taking different values ​​of i and subtracting each pair of equations, we obtain a system of equations.

[0024] ;

[0025] system of equations, transformed into ;

[0026] Solving the above system of equations, since the system consisting of three moving CNC positioners is a redundant drive mechanism with a total of nine axes, three of them are selected as follower axes. Using the above system of equations, the complete equation can be obtained. , , Substituting these values ​​into the previous equation yields the values ​​of R and t, which are then used to calculate... The value of .

[0027] Furthermore, in steps 2-3, the homogeneous transformation matrix of the target point or the entire wing relative to the initial position is solved using the SVD method, specifically including:

[0028] Based on the spatial coordinates of the target point on the wing side obtained in the initial state And the spatial coordinates of the target points on the wing side at m positions. ; Calculate the target's initial position and centroid coordinates as follows The centroid of the target point at any location is ;

[0029] Centralized spatial coordinate value set: , Construct the covariance matrix: ;

[0030] Perform SVD decomposition on the covariance matrix: Where U and V are orthogonal matrices, and Σ is a singular value diagonal matrix; the optimal rotation matrix is ​​obtained as: ,like , revised to , j=1...m;

[0031] Estimating the translation vector: based on the centroid relation have to ;

[0032] Then the homogeneous transformation matrix of the target point or the entire wing relative to the initial position .

[0033] Furthermore, solving for the approximate linear equation using multiple sets of homogeneous transformation matrices yields:

[0034] Extracting the rotation axis direction vector from the homogeneous matrix: Solving unit zero space , which is the direction vector of the rotation axis;

[0035] Solve for the equation ,Right now point coordinates ,Pick Obtain a unique solution;

[0036] Then the equation of the straight line in the space of the rotation axis is: .

[0037] Furthermore, in steps 2-5, the midpoint method of the perpendicular bisector is used to find the intersection point of the lines containing the three axes of rotation.

[0038] Furthermore, steps 2-6 utilize the above data to solve for the optimal parameters, specifically including:

[0039] Since the three axes of motion of the POGO cylinder are orthogonal, the coordinate system {P} is established at the center of the sphere and socket. In the initial state, we directly obtain... ;

[0040] Since the coordinates of the ball socket and the ball head in the camera coordinate system {C} are equal in both ways, we obtain the equation:

[0041] ;

[0042] Right now j=1,2…m;

[0043] Solving for the optimal solution using the SVD method To obtain the optimal The value of .

[0044] A vision-based multi-robot kinematics self-calibration system includes:

[0045] The kinematic model building unit is used to build a kinematic model for the docking of multiple robot systems.

[0046] The docking parameter solving unit, based on the kinematic model, calibrates the docking parameters of a multi-robot system, including:

[0047] A set of spatial coordinate data of the target point on the wing side is obtained through a multi-view vision measurement system. The homogeneous coordinate transformation matrix from the coordinate system of the i-th target point on the wing side to the coordinate system of the interface on the wing side is determined. The spatial coordinate values ​​of the target point on the wing side are obtained in the initial state, as well as the position of each motion axis of the mobile CNC positioner.

[0048] Control the movement of the Z-axis of the i-th movable CNC positioner, while keeping the other two movable CNC positioners fixed. Record the position of each axis of the i-th movable CNC positioner and the spatial coordinates of the target point on the wing side at m positions. Control the movable CNC positioner to return to the initial position, replace the moving movable CNC positioner, and repeat this step until all movable CNC positioners have completed their execution.

[0049] Using the spatial coordinates of the target point on the wing side in the initial state and the spatial coordinates of the target point on the wing side at m positions, the homogeneous transformation matrix of the target point or the entire wing relative to the initial position is solved by the SVD method.

[0050] The approximate equation of the straight line is solved by using multiple sets of homogeneous transformation matrices;

[0051] Find the intersection point of the lines containing the three rotation axes;

[0052] Use the above data to solve for the other parameters.

[0053] A computer storage medium storing an executable program, the executable program being executed by a processor to implement the steps of the multi-robot kinematic self-calibration method.

[0054] Compared with the prior art, the beneficial effects of this invention are as follows: This invention selects one of the POGO pillars, controls its Z-axis movement, and performs force-following movement on the X and Y axes. The axes of the other two POGO pillars are fixed. The motion of each axis and the spatial coordinates of the target point at the wing tip are recorded at m positions. The rotation axis of the wing under the above motion is calculated by the SVD method, which is the line connecting the centers of the ball heads of the two fixed POGO pillars. Other POGO pillars are selected for movement, and the above steps are repeated to obtain three sets of equations for the line connecting the centers of the ball heads of the POGO pillars. By solving the intersection of the three sets of straight lines, three ball head coordinate values ​​can be obtained. Through the above multiple sets of motion data, an optimal model of kinematic parameters is established and solved to obtain the calibrated kinematic parameters. This invention can improve the execution accuracy and attitude adjustment docking efficiency of existing equipment, and has the characteristics of fast calibration and automation. It is suitable for flexible assembly of aircraft wings and fuselages. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a schematic diagram of the composition of a multi-robot system for aircraft wing-body docking provided in an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the structure of the mobile CNC positioner provided in an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of the wing structure provided in an embodiment of the present invention;

[0059] Figure 4 A schematic flowchart of the calibration method provided in an embodiment of the present invention;

[0060] Figure 5 This is a detailed flowchart illustrating the calibration method provided in an embodiment of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] This invention provides a method such as Figure 1The multi-robot system for aircraft wing-body docking shown includes: a mobile CNC positioner 1, a wing 2, a multi-view vision measurement system 3, wingtip visual targets 4, fuselage side visual targets 5, a fuselage 6, and a support 7. The multi-robot system consists of three mobile CNC positioners (POGOs), each with three degrees of freedom. The ball joint at the top of each POGO connects to the ball joint under the wing, forming a ball joint. The combination of the three POGOs enables six-degree-of-freedom wing pose adjustment. Several wingtip visual targets 4 are fixed at the wingtip interface, and their relative positions are ensured through machining. Similarly, several wingtip visual targets 4 are fixed at the wingtip interface, and their relative positions are ensured through machining. The multi-view vision measurement system 3 is located above the fuselage and wings. Its field of view covers the visual target point 4 at the wingtip and the visual target point 5 on the side of the fuselage. By using the coordinates of the visual target point at the wingtip and the visual target point on the side of the fuselage fed back by the multi-view vision system, the relative pose relationship between the wings and the fuselage can be calculated, thereby adjusting the pose.

[0063] The mobile CNC positioner, such as Figure 2 As shown, the system consists of an AGV1-1, a 3-axis CNC positioner 1-2, a force sensor 1-3, a ball joint 1-4, and a recognition camera 1-5. The AGV has omnidirectional mobility and can autonomously navigate or be remotely controlled to the loading area under the wing. This loading area is the AGV's parking position, and the CNC positioner's travel covers this area, ensuring that the positioner can autonomously connect to the wing. The 3-axis CNC positioner achieves 3-axis CNC motion through a motion controller fixed to the upper surface of the AGV. A 3D force sensor is mounted on the upper part of the 3-axis positioner, which identifies the forces acting on the positioner for force-following control. A ball joint is mounted above the 3D force sensor, which, in conjunction with a ball joint under the wing, enables the connection between the CNC positioner and the wing.

[0064] The wing consists of a wing body 2-1, a wing interface 2-2, a ball joint 2-3, and an identification target 2-4, such as Figure 3 As shown. The ball joint is mounted under the wing and connected to the wing via a mechanical interface. The positional relationship of the ball joint relative to the wing is ensured by machining. The calibration target is fixed to the ball joint using tooling, and the relative positional relationship between the ball joint and the calibration target can be determined by the machining accuracy.

[0065] This invention proposes a vision-based multi-robot kinematic self-calibration method. It uses a recognition camera to photograph the target beneath the wing, establishing the relationship between the CNC positioners and the wing coordinate system. Due to the relative positional relationships between the targets beneath the wing, the poses of the three CNC positioners can be roughly calculated. A kinematic model of the entire attitude adjustment mechanism is established, where the unknown kinematic parameters are mainly the pose parameters of the three POGO pillars, and the pose parameters of the three ball heads and the wingtip. The spatial coordinates of the target point at the wingtip are measured using a multi-view vision measurement system. One POGO pillar is selected, and its Z-axis movement is controlled, while the X and Y axes follow the force. The axes of the other two POGO pillars are fixed. The motion of each axis and the spatial coordinates of the target point at the wingtip are recorded at n positions. The rotation axis of the wing can be calculated using the SVD method, which is the line connecting the centers of the ball heads of the two fixed POGO pillars. Other POGO pillars are selected and moved, and the above steps are repeated to obtain the equations of the lines connecting the centers of the ball heads of the three POGO pillars. By solving for the intersection of the three sets of lines, the coordinate values ​​of the three ball heads can be obtained. Using the above sets of motion data, an optimization model for the kinematic parameters is established and solved to obtain the calibrated kinematic parameters. The workflow is as follows: Figure 4 and Figure 5 As shown. Specifically includes:

[0066] First, a kinematic model of the entire docking mechanism is established. The unknown kinematic parameters are mainly the pose parameters of the three POGO pillars, and the pose parameters of the three ball joints and the wing tip.

[0067] The coordinate system is defined as follows:

[0068] After the multi-view vision measurement system is set up, a camera coordinate system {C} is established, and the output three-dimensional coordinates of the target points are all values ​​in the camera coordinate system. A fuselage coordinate system {F} is established at the geometric center of the fuselage-side interface, as follows. Figure 1 As shown. A wing coordinate system {W} is established at the geometric center of the wing-side interface, where the center coordinate system of the k-th wing-side target point is {V}. k}, the coordinate system of the center of the i-th winglet under the ball joint {B i},like Figure 3 As shown. A coordinate system {P} is established at the center of the sphere-socket joint of the i-th movable POGO column. i},like Figure 2 As shown.

[0069] First, let the homogeneous coordinate transformation matrix from the camera coordinate system {C} to the i-th POGO cylinder base coordinate system {Gi} in the multi-view vision measurement system be: ,in Let be the rotation transformation matrix from coordinate system {C} to coordinate system {Gi}. Euler angles for rotational transformation, Let be the translation vector from coordinate system {C} to scale system {Pi}. The homogeneous coordinate transformation matrix from the i-th POGO cylinder base coordinate system {Gi} to the sphere-and-socket coordinate system {Pi} is: ,in Let be the translation vector from coordinate system {Gi} to standard system {Pi}. Since The driving force of the POGO column in three directions can be directly obtained through the control system.

[0070] Let the transformation matrix from coordinate system {Pi} to coordinate system {Bi} be... The unknown is Euler angles The transformation matrix from coordinate system {Bi} to coordinate system {Wi} is: ,in For coordinate system {B i} to coordinate system {W i The translation vector of}.

[0071] Based on the above definitions, the homogeneous transformation matrix from coordinate system {C} to coordinate system {Wi} can be obtained as follows:

[0072] The condition holds true for i = 1, 2, 3, and i takes any of the three values.

[0073] By taking different values ​​for i, we obtain the system of equations. ,available .

[0074] By taking different values ​​of i and subtracting the equations pairwise, we obtain a system of equations:

[0075] ;

[0076] The system of equations can be transformed into: ;

[0077] Solving the above system of equations, since the system composed of POGO pillars is a redundant drive mechanism with a total of 9 axes, 3 of them are selected as follower axes. The complete equations can then be obtained. , , Substituting these values ​​into the previous equation yields the values ​​of R and t, from which the following can be calculated. The value of .

[0078] Based on the above kinematic model, a calibration model for the mechanism parameters is established.

[0079] In a known set , , Given the values ​​of R and t, and the measured values ​​of R and t, solve for a set of structural dimensional parameters. The goal is to find values ​​for (i=1,2,3) such that the theoretical values ​​R* and t* obtained through the kinematic model are closest to the measured values ​​R and t under these values. Essentially, this is an optimization problem. Due to the large number of parameters, the following parameter optimization method is proposed to reduce the difficulty of the optimization solution.

[0080] Step 1: Acquire a set of spatial coordinate data of the target points on the wing side using a multi-view vision system. Based on the design drawings of the wing-side interface, the coordinate system {V} of the k-th wing-side target point can be directly obtained. k Homogeneous coordinate transformation matrix from the wing side interface coordinate system {W} to the wing side interface coordinate system {W} ,like Figure 3 As shown.

[0081] The spatial coordinates of the target point on the wing side are obtained in the initial state. and the position of each motion axis .

[0082] Step 2: Control the Z-axis movement of the i-th POGO column, while keeping the other two POGO columns fixed, and record the position of the i-th POGO column along each axis at m positions. and the spatial coordinates of the target point on the wing side Then, control the POGO column to return to its initial position, replace the moving POGO column, and repeat the above steps until all POGO columns have completed their operations.

[0083] Step 3: Utilize as well as The homogeneous transformation matrix of the target point or the entire wing relative to the initial position is solved using the SVD method. The solution process is as follows.

[0084] First, calculate the centroid of the point set: the initial target position's centroid is... The centroid of the target point at any location is ;

[0085] Centralized point set: , At this point, the relationship is simplified to (Rotation only);

[0086] Construct the covariance matrix: ;

[0087] SVD decomposition: , where U and V are orthogonal matrices, and Σ is a singular value diagonal matrix.

[0088] The optimal rotation matrix is: ,like , revised to ;

[0089] Estimating the translation vector: based on the centroid relation achievable ;

[0090] Then the homogeneous transformation matrix .

[0091] Step 4: Solve for the approximate straight line equation by using multiple sets of homogeneous transformation matrices. The solution process is as follows.

[0092] Extracting the rotation axis direction vector from the homogeneous matrix: Solving unit zero space , which is the direction vector of the rotation axis;

[0093] Find a point on a straight line: solve for the equation. Right now point coordinates Since there are infinitely many solutions, take This will yield a unique solution.

[0094] Then the equation of the straight line in the space of the rotation axis is: .

[0095] Step 5: Find the intersection point of the three axes of rotation. Use the midpoint method of the perpendicular bisector. Taking the intersection point B1 of lines l2 and l3 as an example, the solution process is as follows.

[0096] Calculate intermediate quantities: ;

[0097] ;

[0098] ;

[0099] Calculate the intersection point of the common perpendicular with the two lines:

[0100] Intersection point of line 2: ;

[0101] Intersection point of line 3: ;

[0102] Midpoint coordinates: ;

[0103] Repeat the above steps to obtain .

[0104] Step 6: Solve for the optimal solution using the above data. and .

[0105] Since the three axes of motion of the POGO cylinder are orthogonal, the coordinate system {P} can be established at the center of the sphere and socket. In the initial state, we can directly obtain... We only need to find the optimal solution. ;

[0106] The coordinates of the ball socket and the ball head in the camera coordinate system {C} are equal in both ways, and the equation can be obtained.

[0107] ;

[0108] Right now j=1,2…m;

[0109] The optimal solution can be obtained using the SVD method. The optimal result can then be obtained. The value of .

[0110] The present invention also provides a vision-based multi-robot kinematics self-calibration system, comprising:

[0111] The kinematic model building unit is used to build a kinematic model for the docking of multiple robot systems.

[0112] The docking parameter solving unit, based on the kinematic model, calibrates the docking parameters of a multi-robot system, including:

[0113] A set of spatial coordinate data of the target point on the wing side is obtained through a multi-view vision measurement system. The homogeneous coordinate transformation matrix from the coordinate system of the i-th target point on the wing side to the coordinate system of the interface on the wing side is determined. The spatial coordinate values ​​of the target point on the wing side are obtained in the initial state, as well as the position of each motion axis of the mobile CNC positioner.

[0114] Control the movement of the Z-axis of the i-th movable CNC positioner, while keeping the other two movable CNC positioners fixed. Record the position of each axis of the i-th movable CNC positioner and the spatial coordinates of the target point on the wing side at m positions. Control the movable CNC positioner to return to the initial position, replace the moving movable CNC positioner, and repeat this step until all movable CNC positioners have completed their execution.

[0115] Using the spatial coordinates of the target point on the wing side in the initial state and the spatial coordinates of the target point on the wing side at m positions, the homogeneous transformation matrix of the target point or the entire wing relative to the initial position is solved by the SVD method.

[0116] The approximate equation of the straight line is solved by using multiple sets of homogeneous transformation matrices;

[0117] Find the intersection point of the lines containing the three rotation axes;

[0118] Use the above data to solve for the other parameters.

[0119] A computer storage medium storing an executable program, the executable program being executed by a processor to implement the steps of the multi-robot kinematic self-calibration method.

[0120] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0121] Obviously, those skilled in the art can make various modifications and variations to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Therefore, if these modifications and variations to the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A multi-robot kinematic self-calibration method based on vision measurement for a multi-robot system of aircraft wing-body docking, the multi-robot system comprising three mobile CNC positioners, a wing, a multi-view vision measurement system, a visual target point at the wingtip, a visual target point on the fuselage side, a fuselage, and a support, characterized in that... include: Step 1: Establish a kinematic model for the docking of multiple robot systems; Step 2, based on the kinematic model, calibrate the docking parameters of the multi-robot system; including: Step 2-1: Obtain a set of spatial coordinate data of the target point on the wing side through a multi-view vision measurement system, determine the homogeneous coordinate transformation matrix from the coordinate system of the i-th target point on the wing side to the coordinate system of the interface on the wing side, obtain the spatial coordinate values ​​of the target point on the wing side in the initial state, and the position of each motion axis of the mobile CNC positioner. Step 2-2: Control the Z-axis movement of the i-th movable CNC positioner, while keeping the other two movable CNC positioners fixed. Record the position of each axis of the i-th movable CNC positioner and the spatial coordinates of the target point on the wing side at m positions. Control the movable CNC positioner to return to the initial position, replace the moving movable CNC positioner, and repeat this step until all movable CNC positioners have completed their execution. Steps 2-3: Using the spatial coordinates of the target point on the wing side in the initial state and the spatial coordinates of the target point on the wing side at m positions, solve the homogeneous transformation matrix of the target point or the entire wing relative to the initial position using the SVD method. Steps 2-4 involve solving for the approximate equation of the straight line using multiple sets of homogeneous transformation matrices. Steps 2-5: Find the intersection points of the lines containing the three rotation axes; Steps 2-6: Use the above data to solve for other parameters.

2. The multi-robot kinematics self-calibration method based on vision measurement according to claim 1, characterized in that, Establish a kinematic model for multi-robot system docking, including: In a multi-view vision measurement system, a camera coordinate system {C} is established, where the output target point's 3D coordinates are all values ​​in the camera coordinate system. A fuselage coordinate system {F} is established based on the geometric center of the fuselage-side interface, and a wing coordinate system {W} is established based on the geometric center of the wing-side interface. A coordinate system {V} is established at the center of the k-th target point on the wing side. k Establish a coordinate system {B} at the center of the i-th ball joint under the wing. i Establish a coordinate system {P} at the center of the ball socket of the i-th moving CNC positioner. i A coordinate system {G} is established at the center of the base of the i-th mobile CNC positioner. i }; where i=1,2,3; k=1...n, and n is the number of target points; Let the camera coordinate system {C} in the multi-view vision measurement system be connected to the base coordinate system {G} of the i-th moving CNC positioner. i The homogeneous coordinate transformation matrix of} is ,in From coordinate system {C} to coordinate system {G} i The rotation transformation matrix of} Euler angles for rotational transformation, From coordinate system {C} to scale system {P} i Translation vector of}; base coordinate system {G} i } to the spherical-socket coordinate system {P i The homogeneous coordinate transformation matrix of} is ,in For coordinate system {G i } to the standard system {P i The translation vector of}; Let the transformation matrix from coordinate system {Pi} to coordinate system {Bi} be... The unknown is Euler angles. Coordinate system {B i } to coordinate system {W i The transformation matrix of} is ,in Let coordinate system {B i } to coordinate system {W i The translation vector of}; Obtain coordinate system {C} to coordinate system {W} i The homogeneous transformation matrix of} is: ； The system of equations is obtained by rearranging the equations. ; get ; By taking different values ​​of i and subtracting each pair of equations, we obtain a system of equations. ； system of equations, transformed into ; Solving the above system of equations, since the system consisting of three moving CNC positioners is a redundant drive mechanism with a total of nine axes, three of them are selected as follower axes. Using the above system of equations, the complete equation can be obtained. , , Substituting these values ​​into the previous equation yields the values ​​of R and t, which are then used to calculate... The value of .

3. The multi-robot kinematics self-calibration method based on vision measurement according to claim 2, characterized in that, In steps 2-3, the homogeneous transformation matrix of the target point or the entire wing relative to the initial position is solved using the SVD method, specifically including: Based on the spatial coordinates of the target point on the wing side obtained in the initial state And the spatial coordinates of the target points on the wing side at m positions. ; Calculate the target's initial position and centroid coordinates as follows The centroid of the target point at any location is ; Centralized spatial coordinate value set: , Construct the covariance matrix: ; Perform SVD decomposition on the covariance matrix: Where U and V are orthogonal matrices, and Σ is a singular value diagonal matrix; the optimal rotation matrix is ​​obtained as: ,like , revised to , j=1...m; Estimating the translation vector: based on the centroid relation have to ; Then the homogeneous transformation matrix of the target point or the entire wing relative to the initial position .

4. The multi-robot kinematics self-calibration method based on vision measurement according to claim 3, characterized in that, The approximate linear equation obtained by solving multiple sets of homogeneous transformation matrices is: Extracting the rotation axis direction vector from the homogeneous matrix: Solving unit zero space , which is the direction vector of the rotation axis; Solve for the equation ,Right now Point coordinates ,Pick Obtain a unique solution; Then the equation of the straight line in the space of the rotation axis is: .

5. The multi-robot kinematics self-calibration method based on vision measurement according to claim 4, characterized in that, Steps 2-5 use the midpoint method of the perpendicular bisector to find the intersection of the lines containing the three axes of rotation.

6. The multi-robot kinematics self-calibration method based on vision measurement according to claim 5, characterized in that, Steps 2-6 use the above data to solve for the optimal parameters, specifically including: Since the three axes of motion of the POGO cylinder are orthogonal, the coordinate system {P} is established at the center of the sphere and socket. In the initial state, we directly obtain... ; Since the coordinates of the ball socket and the ball head in the camera coordinate system {C} are equal in both ways, we obtain the equation: ； Right now j=1,2…m; Solving for the optimal solution using the SVD method To obtain the optimal The value of .

7. A vision-based multi-robot kinematics self-calibration system for implementing the method of any one of claims 1-6, characterized in that, include: The kinematic model building unit is used to build a kinematic model for the docking of multiple robot systems. The docking parameter solving unit, based on the kinematic model, calibrates the docking parameters of a multi-robot system, including: A set of spatial coordinate data of the target point on the wing side is obtained through a multi-view vision measurement system. The homogeneous coordinate transformation matrix from the coordinate system of the i-th target point on the wing side to the coordinate system of the interface on the wing side is determined. The spatial coordinate values ​​of the target point on the wing side are obtained in the initial state, as well as the position of each motion axis of the mobile CNC positioner. Control the movement of the Z-axis of the i-th movable CNC positioner, while keeping the other two movable CNC positioners fixed. Record the position of each axis of the i-th movable CNC positioner and the spatial coordinates of the target point on the wing side at m positions. Control the movable CNC positioner to return to the initial position, replace the moving movable CNC positioner, and repeat this step until all movable CNC positioners have completed their execution. Using the spatial coordinates of the target point on the wing side in the initial state and the spatial coordinates of the target point on the wing side at m positions, the homogeneous transformation matrix of the target point or the entire wing relative to the initial position is solved by the SVD method. The approximate equation of the straight line is solved by using multiple sets of homogeneous transformation matrices; Find the intersection point of the lines containing the three rotation axes; Use the above data to solve for the other parameters.

8. A computer storage medium, characterized in that, The computer storage medium stores an executable program, which is executed by a processor to implement the steps of the multi-robot kinematic self-calibration method according to any one of claims 1-6.

Images