High-precision collaborative predictive control method for double transfer robots

Abstract

The invention discloses a high-precision collaborative predictive control method for double transfer robots. The method comprises the steps that a hardware platform of a collaborative control system for the double transfer robots is built; measuring and establishing a global coordinate system and a local coordinate system of each component, and constructing a coordinate system conversion model; according to the position relation mapping, coarse positioning of a transfer platform in the transfer robot system and solving of a machining track of an end effector are carried out; constructing a double-robot cross coupling collaborative predictive control model, and calculating the positioning error of each transfer robot system and the collaborative trajectory error between the double transfer robot systems; solving a compensation value and predicting an output pose at the next moment; the output poses are solved into robot joint angles through an inverse solution device and sent to corresponding motor drivers, and cooperative motion control over the double-transfer robot is achieved. According to the method provided by the invention, high-precision and high-quality double-robot collaborative operation of a large workpiece in-situ machining task is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of dual-robot collaborative control, and in particular relates to a high-precision collaborative predictive control method for dual transfer robots. Background Art

[0002] As workpieces become increasingly larger, their dimensions exceed the capabilities of traditional machine tools. Furthermore, the positional correlations between surface features on large workpieces necessitate high relative positioning accuracy, posing significant challenges to existing machining technologies. In recent years, in-situ operations, using mobile robots as manufacturing units, have found practical applications in areas such as grinding, welding, painting, and assembly.

[0003] Because a single mobile robot needs to move across multiple processing areas when machining multiple related features, the coordinate relationship between the robot and the workpiece changes multiple times, resulting in multiple sets of cumulative random errors. Compared to single mobile robot operations, dual mobile robots are set up in a master-slave robot mode. During collaborative machining, the slave robots can be synchronously controlled according to the master robot's trajectory, ensuring that related features are machined simultaneously and with high-precision relative position relationships. In addition, due to the different centers of gravity of the robots in different postures, excessive robot overhang can easily cause deformation of the robot base mounting plate, affecting the robot's end-of-line posture accuracy. Therefore, a high-precision collaborative predictive control method for dual transfer robots is required.

[0004] In the existing technology, there are methods that use industrial cameras and lasers to locate the weld seam of the welding object based on the data information of the welding object, and multiple robots cooperate with each other to realize the flipping and welding operations of the welding object in the same workstation area, thereby improving the efficiency and accuracy of the welding industry; there is also a method that analyzes the movement strategy of the laser welder with the highest welding efficiency under the premise of minimizing the mutual influence of synchronous welding, and corrects the welding route in real time through the camera, thereby solving the problems of poor weld formation, stress concentration and deformation caused by the close distance between the two welding operations.

[0005] However, the above methods are all applied in the welding field, and have low requirements for robot positioning accuracy, which cannot meet the accuracy requirements in fields such as hole making, milling and assembly; and the methods focus on hardware architecture optimization and control timing scheduling strategies, and have not explored the collaborative accuracy of multiple robots under dynamic load conditions, and the correlation between multiple robots is relatively weak.

[0006] Existing methods establish a mathematical model for multi-robot collaborative handling by using the kinematic model of the gripping point and the target's center of mass as constraints, thereby achieving synchronized motion of the robot end points. However, this method does not consider the positioning accuracy of the robot end points. The coordinated motion of multiple robots may result in cumulative errors, leading to asynchrony during the multi-robot collaborative motion, which in turn affects the quality of the multi-robot collaborative work. This makes it unsuitable for tasks requiring high precision. Summary of the Invention

[0007] In response to the problems existing in the above-mentioned prior art, the present invention proposes a high-precision collaborative predictive control method for dual transfer robots to achieve synchronous convergence of the trajectory error of the single-side robot and the collaborative trajectory error of the dual robots, thereby realizing high-precision and high-quality dual-robot collaborative operation.

[0008] In order to achieve the above technical objectives, the present invention provides the following technical solutions:

[0009] A high-precision collaborative predictive control method for dual transfer robots, which specifically includes the following steps:

[0010] S1. Build a dual-transfer robot collaborative control system hardware platform, including two transfer robot systems, a binocular camera as a visual servo device, large workpieces to be processed, and an industrial computer equipped with the dual-transfer robot collaborative control system; each transfer robot system includes a six-degree-of-freedom robot, a transfer platform, and an end effector;

[0011] S2. Use binocular cameras to measure and establish a global coordinate system and local coordinate systems for each component, and construct a coordinate system transformation model to determine the positional relationship mapping between the large workpiece to be processed, the six-degree-of-freedom robot, the end effector, and the transfer platform;

[0012] S3. Based on the position relationship mapping, the coarse positioning of the transfer platform and the solution of the end effector processing trajectory are performed;

[0013] S4. Construct a dual-robot cross-coupling collaborative predictive control model to calculate the positioning error of each transfer robot system at the current moment and the collaborative trajectory error between the transfer robot systems, and determine whether the two transfer robot systems have achieved collaboration; then solve the compensation value at the current moment and predict the output position of the end effector at the next moment;

[0014] S5. Calculate the joint angles of each six-degree-of-freedom robot under the output posture of the end execution at the next moment through the corresponding robot inverse solver; send the joint angles to the corresponding motor drivers to realize the coordinated motion control of the dual transfer robots.

[0015] Furthermore, in step S1:

[0016] In each transfer robot system: the six-degree-of-freedom robot is fixed on the transfer platform as a whole, and the end effector is installed at the end of the sixth joint of the six-degree-of-freedom robot;

[0017] The dual-transfer robot collaborative control system integrates functional modules including binocular camera data reading, coordinate system conversion model, tool path planning, dual-robot cross-coupling collaborative predictive control model, robot inverse solver, and control modules for the operation control of six-degree-of-freedom robots, transfer platforms and end effectors.

[0018] Furthermore, step S2 specifically includes:

[0019] S21, establish the target feature coordinate system of the large workpiece to be processed {W i}、Six-DOF robot base coordinate system {B i}、End effector coordinate system {E i}、Transfer platform coordinate system {A i}, binocular camera coordinate system {C} and global coordinate system {G}; where i = 1, 2, representing the first and second six-degree-of-freedom robots;

[0020] S22, through the coordinate system conversion model, obtain the target feature coordinate system {W i}Transformation matrix relative to the global coordinate system {G} Transfer platform coordinate system {A i}Transformation matrix relative to the global coordinate system {G} Transformation matrix of the binocular camera coordinate system {C} relative to the global coordinate system {G}

[0021] S23, then obtain the coordinate system of the transfer platform {A i}Relative to the target feature coordinate system {W i}'s transformation matrix End effector coordinate system {E i}Transformation matrix relative to the global coordinate system {G} End effector coordinate system {E i}Relative to the target feature coordinate system {W i}'s transformation matrix

[0022] Furthermore, step S3 specifically includes:

[0023] S31. Determine the target position information of the large workpiece to be processed based on the transformation matrix of the transfer platform coordinate system relative to the target feature coordinate system; based on this information, the dual transfer robot collaborative control system drives the transfer platform to perform station conversion, and activates the transfer platform support mechanism after reaching the target area to achieve coarse positioning of the transfer platform;

[0024] S32. Then, according to the position information of the current end effector coordinate system relative to the target feature coordinate system, the machining trajectory is solved by the tool path planner.

[0025] Furthermore, step S4 specifically includes:

[0026] Step S4 specifically includes:

[0027] S41, at each moment, the actual position of the end effector is monitored and fed back in real time by the binocular camera, and the position information of the end effector at that moment is extracted from the processing trajectory as the ideal position of the end effector, thereby obtaining the positioning error matrix X of the single transfer robot system l , the formula is:

[0028]

[0029] Among them, X 1_l =[x i ,y i ,z i ,α i ,β i ,γ i ] T is the pose error matrix composed of the single robot translation error and pose error, For X 1_l Differentiation with respect to time; l is 1 and 2 respectively represent the first and second transfer robot systems;

[0030] S42. Determine the positioning error of each transfer robot system based on the positioning error model. Then, using one of the dual transfer robots as a master robot and the other as a slave robot, determine the collaborative trajectory error and cross-coupling error between the master and slave robots based on the positioning error of the single transfer robot system.

[0031] S43. Construct a state space model of the dual-robot collaborative control system. The formula is expressed as:

[0032]

[0033] Among them, x(k) and x(k+1) are the collaborative trajectory error states at time k+1 and time k respectively. is the control input, is the controlled output, is the external disturbance, A is the state transfer matrix, B u 、B d are the control input parameter matrix and the external disturbance parameter matrix, C c is the observation matrix;

[0034] S44. Based on the state space model of the dual-robot collaborative control system and combined with the current state input, predict the controlled output for multiple steps in the future, with the goal of minimizing the cross-coupling error, and solve the optimal control input increment as the compensation value and the output posture of the end effector at the next moment.

[0035] More specifically, step S42 is as follows:

[0036] According to the positioning error model of a single transfer robot system, the positioning error e of the master and slave robots is obtained. r_1 、e r_2 , that is, e r_1 =X1,e r_2 =X2; then define the collaborative trajectory error between the master and slave robots as:

[0037] ε r_1 =e r_1 -e r_2 ,ε r_2 =e r_2 -e r_1 ;

[0038] Among them, ε r_1 With ε r_2 are the collaborative trajectory errors of the master and slave robots respectively. When ε r_1 = 0 and ε r_2 =0 means that the two robots realize cooperative control;

[0039] The error matrix of the dual-robot collaborative trajectory is denoted as Δ; the cross-coupling error between the two transfer robot systems is denoted as Γ; the formulas for Δ and Γ are expressed as:

[0040]

[0041] Γ=[e c_1 ,e c_2 ] T =E+μΔ=(I+μT)E;

[0042] Among them, e c_1 With e c_2 are the cross-coupling errors of the master and slave robots, respectively; μ is a proportional factor used to control the cooperation between the two robots.

[0043] More specifically, step S44 includes:

[0044] S441. Calculate the predicted increment of the collaborative trajectory error state variable at time k to time k+1, Δx(k+1|k). The formula is:

[0045] Δx(k+1|k)=AΔx(k)+Bu Δu(k)+B d Δd(k);

[0046] in, is the collaborative trajectory error state increment, is the control input increment, Δd(k) is the external disturbance increment;

[0047] S442, by recursion, obtain the predicted value of the coordinated trajectory error state at time k for time k+p, and then substitute it back into the state space model to obtain the predicted value y of the controlled output at time k for time k+p c (k+p|k), ​​the formula is expressed as:

[0048]

[0049] S443, perform rolling optimization on the predicted value of the controlled output; when there is a control input increment, optimize y by proportional relationship and superposition principle. c (k+p|k) is simplified to obtain the predicted value of the controlled output at the future moment under the action of the one-step control input increment. The formula is:

[0050]

[0051] in, is the predicted value of the controlled output at the future moment when no control input increment is added; a i is the unit step response coefficient, N is the optimization time domain, that is, the number of optimization steps in the future; Δu(k)=u(k)-u(k-1) represents the control input increment at time k relative to time k-1;

[0052] Then recursively obtain the predicted value of the controlled output at the future moment under the M-step control input increment

[0053] S444, for Define the performance evaluation index minJ(k); the formula is expressed as:

[0054]

[0055] Among them, q i 、r j are the error weighting factor and control constraint factor, which respectively reflect the degree of suppression of the collaborative trajectory error and the constraint strength of the control input increment; w(k+i) is the expected value of the control input increment determined at each time point k for the controlled object at the next P time points; M is the control horizon, that is, the actual number of control adjustment steps and M≤P≤N;

[0056] remember is a matrix composed of the predicted values ​​of the controlled output in the next P steps under the control input increment, is the future P steps without control input increment The matrix composed of the controlled output prediction values; satisfy:

[0057]

[0058] Where Δu M (k) is the M-step control input increment matrix, and the j-th step control input increment is expressed as Δu j (k)=U j T Δu M (k); U j is a row vector whose jth entry is 1 and the rest are 0;

[0059]

[0060] according to The performance evaluation index minJ(k) is simplified by the relationship between

[0061]

[0062] Among them, w P (k) = [w(k+1)Lw(k+P)] T ; Q, R are the error weight matrix and control constraint matrix, Q = diag (q1, L, q P ), R=diag(r1,L,r M );

[0063] At time k, w P (k), All are known, then according to the extreme value necessary condition dJ(k) / dΔu M (k) = 0, find the optimal M-step control input increment matrix that minimizes J(k), denoted as Δu M (k)', the formula is expressed as:

[0064]

[0065] That is, the compensation value in the future control time domain of size M is obtained;

[0066] S445, at time k, Δu M (k)' is the cross-coupling error applied to the lth robot, then the predicted output value y under the action of the optimal control input increment is N1 (k) is:

[0067] y N1 (k) = yN0 (k)+a1'Δu1(k)';

[0068] Among them, y N0 (k) is the controlled output when no control input increment is added; Δu1(k)' is Δu M The first step in (k)' is the control input increment, and a1' is the corresponding unit step response coefficient;

[0069] The actual output y(k+1) at time k+1 is compared with the predicted output value of the controlled object at time k+1 under the action of the one-step control input increment. By comparison, we can get the prediction error e(k+1) at time k+1; the expression is:

[0070]

[0071] Combined with the calculated optimal control input increment matrix, the prediction error is heuristically corrected to obtain the predicted value of the controlled output at time k+1 under the corrected optimal control input increment. for:

[0072]

[0073] That is, the output pose of the end effector at time k+1 is obtained; wherein the correction vector h is an N-dimensional vector consisting of weight coefficients h = [h1Lh N ] T ;

[0074] The subsequent time steps continue to perform online iterations using the rolling optimization strategy to obtain the output pose for each time step in turn.

[0075] Based on the above technical solution, the present invention has at least the following beneficial effects:

[0076] 1) The high-precision collaborative predictive control method for dual transfer robots proposed in this invention plans the positions of the dual transfer robots, solves the collaborative processing trajectories of the dual robots, and realizes that the dual transfer robots maintain a certain posture association during operation, thereby collaboratively processing associated features with relative positioning accuracy requirements.

[0077] 2) In order to solve the problem that the transfer robot system exceeds the monitoring range of the binocular camera after the workstation is converted, the present invention proposes that the binocular camera can be placed on the transfer platform and move with the transfer platform. By constructing a global coordinate system to associate the local coordinate systems, the accurate relative positions of the local coordinate systems can be determined.

[0078] 3) The present invention addresses the difficulty of low single robot positioning accuracy and dual robot collaboration accuracy caused by multiple source errors such as robot system error, transfer platform flexibility error and external uncertain disturbance error, and constructs a dual robot cross-coupling collaborative predictive control model; this model correlates the error between the actual posture and theoretical posture of each robot at the current moment in real time to obtain the collaborative trajectory error of the dual robots, and realizes the synchronous convergence of the single robot trajectory error and the dual robot collaborative trajectory error, thereby compensating for the robot posture at the next moment and improving the dual robot collaborative control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 This is a flow chart of a high-precision collaborative predictive control method for dual transfer robots of the present invention;

[0080] Figure 2 This is a schematic diagram of the hardware layout and coordinate system distribution of the dual-transfer robot processing scenario;

[0081] Figure 3 Schematic diagram of the control structure of the dual-robot cross-coupling collaborative predictive control model. DETAILED DESCRIPTION

[0082] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, the following Figure 1-3 The present invention is further described in detail with specific implementation methods, so that the application can fully understand how to use technical means to solve technical problems and achieve technical effects and implement them accordingly.

[0083] Those skilled in the art will appreciate that all or part of the steps in the above-mentioned embodiment methods can be accomplished by instructing the relevant hardware through a program. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] This paper proposes a high-precision collaborative predictive control method for dual transfer robots, used in in-situ manufacturing of large workpieces. This method improves the relative positional accuracy between related features on the workpiece's surface while ensuring the stable operation of the transfer robots. This high-precision collaborative control of the dual transfer robots involves setting one robot as the master and the other as a slave. The slave adjusts its trajectory based on the master's trajectory, thereby achieving collaborative operation between the two robots.

[0085] like Figure 1As shown, the high-precision collaborative control method of dual transfer robots proposed in the present invention specifically includes the following steps:

[0086] S1. Build the hardware platform of the dual-transfer robot collaborative control system, such as Figure 2 As shown, it includes two transfer robot systems, a binocular camera as a visual servo device, a large workpiece to be processed and an industrial computer equipped with a dual transfer robot collaborative control system; each transfer robot system includes a six-degree-of-freedom robot, a transfer platform, and an end effector; as a preferred embodiment, in step S1

[0087] In each transfer robot system: the six-degree-of-freedom robot is fixed on the transfer platform as a whole, and the end effector is installed at the end of the sixth joint of the six-degree-of-freedom robot;

[0088] The dual-transfer robot collaborative control system integrates functional modules including binocular camera data reading, coordinate system conversion model, tool path planning, dual-robot cross-coupling collaborative predictive control model, robot inverse solver, and control modules for the operation control of the six-degree-of-freedom robot, transfer platform, and end effector.

[0089] In addition, the binocular camera can be placed in a fixed position or on a transfer platform to move with the transfer robot system to prevent the transfer robot system from exceeding the monitoring range of the binocular vision.

[0090] S2. Use binocular cameras to measure and establish a global coordinate system and local coordinate systems for each component, and construct a coordinate system transformation model to determine the positional relationship mapping between the large workpiece to be processed, the six-degree-of-freedom robot, the end effector, and the transfer platform;

[0091] As a preferred embodiment, step S2 specifically includes:

[0092] S21, establish the target feature coordinate system of the large workpiece to be processed {W i}、Six-DOF robot base coordinate system {B i}、End effector coordinate system {E i}、Transfer platform coordinate system {A i}, binocular camera coordinate system {C} and global coordinate system {G}; where i = 1, 2, representing the first and second six-degree-of-freedom robots;

[0093] S22, through the coordinate system conversion model, obtain the target feature coordinate system {W i}Transformation matrix relative to the global coordinate system {G} Transfer platform coordinate system {A i}Transformation matrix relative to the global coordinate system {G} Transformation matrix of the binocular camera coordinate system {C} relative to the global coordinate system {G}

[0094] S23, then obtain the coordinate system of the transfer platform {A i}Relative to the target feature coordinate system {W i}'s transformation matrix Transformation matrix of the end effector coordinate system {Ei} relative to the global coordinate system {G} The transformation matrix of the end effector coordinate system {Ei} relative to the target feature coordinate system {Wi}

[0095] In currently used robot precision compensation methods, the positions of measurement devices such as cameras are fixed, and the relationships between coordinate systems are also fixed. However, in this invention, the binocular camera can be fixed on a transfer platform and move with the robot's transfer platform, resolving the problem of the robot changing its work area and potentially exceeding its field of view, making it impossible to monitor. By establishing a fixed global coordinate system, other coordinate systems are bound to it, allowing for real-time conversion of the relative positions between coordinate systems.

[0096] S3. Based on the position relationship mapping, the coarse positioning of the transfer platform and the solution of the end effector processing trajectory are performed;

[0097] As a preferred embodiment, step S3 specifically includes:

[0098] S31. Determine the target position information of the large workpiece to be processed based on the transformation matrix of the transfer platform coordinate system relative to the target feature coordinate system; based on this information, the dual transfer robot collaborative control system drives the transfer platform to perform station conversion, and activates the transfer platform support mechanism after reaching the target area to achieve coarse positioning of the transfer platform;

[0099] S32. Then, based on the position information of the current end effector coordinate system relative to the target feature coordinate system, a tool path planner is used to calculate the machining trajectory. In this embodiment, the machining trajectory is calculated based on the current pose information of the end effector relative to the workpiece coordinate system, and the planner used is an existing dedicated software.

[0100] S4. Construct a dual-robot cross-coupling collaborative predictive control model to calculate the positioning error of each transfer robot system at the current moment and the collaborative trajectory error between the transfer robot systems, and determine whether the two transfer robot systems have achieved collaboration; then solve the compensation value at the current moment and predict the output position of the end effector at the next moment;

[0101] As a preferred embodiment, Figure 3 As shown, step S4 specifically includes:

[0102] S41, at each moment, the actual position of the end effector is monitored and fed back in real time by the binocular camera, and the position information of the end effector at that moment is extracted from the processing trajectory as the ideal position of the end effector, thereby obtaining the positioning error matrix X of the single transfer robot system l , the formula is:

[0103]

[0104] Among them, X 1_l =[x i ,y i ,z i ,α i ,β i ,γ i ] T is the pose error matrix composed of the single robot translation error and pose error (i.e. the difference between the actual pose and the ideal pose), For X 1_l Differentiation with respect to time; l is 1 and 2 represents the first and second transfer robot systems respectively; since the application designs a nonlinear dual-robot system, the differential of the posture error matrix with respect to time is also added when considering the positioning error matrix;

[0105] S42. Determine the positioning error of each transfer robot system based on the positioning error model. Then, using one of the dual transfer robots as a master robot and the other as a slave robot, determine the collaborative trajectory error and cross-coupling error between the master and slave robots based on the positioning error of the single transfer robot system.

[0106] More specifically, step S42 is as follows:

[0107] According to the positioning error model of a single transfer robot system, the positioning error e of the master and slave robots is obtained. r_1 、e r_2 , that is, e r_1 =X1,e r_2 =X2; then define the collaborative trajectory error between the master and slave robots as:

[0108] ε r_1 =e r_1 -e r_2 ,ε r_2 =e r_2 -e r_1 ;

[0109] Among them, ε r_1 With ε r_2 are the collaborative trajectory errors of the master and slave robots respectively. When ε r_1 = 0 and ε r_2 =0 means that the two robots realize cooperative control;

[0110] The error matrix of the dual-robot collaborative trajectory is denoted as Δ; the cross-coupling error between the two transfer robot systems is denoted as Γ; the formulas for Δ and Γ are expressed as:

[0111]

[0112] Γ=[e c_1 ,e c_2 ] T =E+μΔ=(I+μT)E;

[0113] Among them, e c_1 With e c_2 are the cross-coupling errors of the master and slave robots, respectively; μ is a proportional factor used to control the cooperation between the two robots.

[0114] S43. Construct a state space model of the dual-robot collaborative control system. This application aims to achieve convergence of the dual-robot collaborative trajectory error. For a nonlinear dual-robot system, the collaborative trajectory error at each moment is used as the state input of the state space model. The formula is expressed as follows:

[0115]

[0116] Among them, x(k) and x(k+1) are the collaborative trajectory error states at time k+1 and time k respectively. is the control input, is the controlled output, is the external disturbance, A is the state transfer matrix, B u 、B d are the control input parameter matrix and the external disturbance parameter matrix, C c is the observation matrix; in addition, in this embodiment, the control input is the input value of the six joints of the six-degree-of-freedom robot, and the control input can be angle, torque, etc. according to the actual situation;

[0117] S44. Based on the state space model of the dual-robot collaborative control system and the current state input, predict the controlled output for multiple steps in the future, with the goal of minimizing the cross-coupling error, and solve the optimal control input increment as the compensation value and the output pose of the end effector at the next moment;

[0118] As a preferred embodiment, step S44 specifically includes:

[0119] S441. Calculate the predicted increment of the collaborative trajectory error state variable at time k to time k+1, Δx(k+1|k). The formula is:

[0120] Δx(k+1|k)=AΔx(k)+B u Δu(k)+B d Δd(k);

[0121] in, is the collaborative trajectory error state increment, is the control input increment, Δd(k) is the external disturbance increment;

[0122] S442, by recursion, obtain the predicted value of the coordinated trajectory error state at time k for time k+p, and then substitute it back into the state space model to obtain the predicted value y of the controlled output at time k for time k+p c (k+p|k), ​​the formula is expressed as:

[0123]

[0124] S443, perform rolling optimization on the predicted value of the controlled output; when there is a control input increment, optimize y by proportional relationship and superposition principle. c (k+p|k) is simplified to obtain the predicted value of the controlled output at the future moment under the action of the one-step control input increment. The formula is:

[0125]

[0126] in, is the predicted value of the controlled output at the future moment when no control input increment is added; a i is the unit step response coefficient, N is the optimization time domain, that is, the number of optimization steps in the future; Δu(k)=u(k)-u(k-1) represents the control input increment at time k relative to time k-1;

[0127] Then recursively obtain the predicted value of the controlled output at the future moment under the M-step control input increment

[0128] S444, in order to achieve the convergence of the master-slave robot positioning error and the synchronous convergence of the collaborative trajectory error of the dual robots, here is Define the performance evaluation index minJ(k); the formula is expressed as:

[0129]

[0130] Among them, q i 、r jare the error weighting factor and control constraint factor, which respectively reflect the degree of suppression of the collaborative trajectory error and the constraint strength of the control input increment; w(k+i) is the expected value of the control input increment determined at each time point k for the controlled object at the next P time points; M is the control horizon, that is, the actual number of control adjustment steps and M≤P≤N;

[0131] remember is a matrix composed of the predicted values ​​of the controlled output in the next P steps under the control input increment, is the future P steps without control input increment The matrix composed of the controlled output prediction values; satisfy:

[0132]

[0133] Where Δu M (k) is the M-step control input increment matrix, and the j-th step control input increment is expressed as Δu j (k)=U j T Δu M (k); U j is a row vector whose jth entry is 1 and the rest are 0;

[0134]

[0135] according to The performance evaluation index minJ(k) is simplified by the relationship between

[0136]

[0137] Among them, w P (k) = [w(k+1)Lw(k+P)] T ; Q, R are the error weight matrix and control constraint matrix, Q = diag (q1, L, q P ), R=diag(r1,L,r M );

[0138] At time k, w P (k), All are known, then according to the extreme value necessary condition dJ(k) / dΔu M (k) = 0, find the optimal M-step control input increment matrix that minimizes J(k), denoted as Δu M (k)', the formula is expressed as:

[0139]

[0140] That is, the compensation value in the future control time domain of size M is obtained;

[0141] S445, at time k, Δu M (k)' is the cross-coupling error applied to the lth robot, then the predicted output value y under the action of the optimal control input increment is N1 (k) is:

[0142] y N1 (k) = y N0 (k)+a1'Δu1(k)';

[0143] Among them, y N0 (k) is the controlled output when no control input increment is added; Δu1(k)' is Δu M The first step in (k)' is the control input increment, and a1' is the corresponding unit step response coefficient;

[0144] The actual output y(k+1) at time k+1 is compared with the predicted output value of the controlled object at time k+1 under the action of the one-step control input increment. By comparison, we can get the prediction error e(k+1) at time k+1; the expression is:

[0145]

[0146] Combined with the calculated optimal control input increment matrix, the prediction error is heuristically corrected to obtain the predicted value of the controlled output at time k+1 under the corrected optimal control input increment. for:

[0147]

[0148] That is, the output pose of the end effector at time k+1 is obtained; wherein the correction vector h is an N-dimensional vector consisting of weight coefficients h = [h1Lh N ] T ;

[0149] The subsequent time steps continue to iterate online using the rolling optimization strategy, and the output pose of each time step is obtained in turn.

[0150] S5. Calculate the joint angles of each six-degree-of-freedom robot under the output posture of the end execution at the next moment through the corresponding robot inverse solver; send the joint angles to the corresponding motor drivers to realize the coordinated motion control of the dual transfer robots.

[0151] In summary, the present invention proposes a dual-robot collaborative operation method for in-situ processing tasks of large workpieces, which can achieve synchronous convergence of single-side robot trajectory error and dual-robot collaborative trajectory error, thereby achieving high-precision and high-quality dual-robot collaborative operation, effectively improving the relative positioning accuracy and processing efficiency between workpiece features, and has practicality and versatility.

[0152] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0153] The logic and / or steps represented in the flowchart or otherwise described herein may be considered, for example, as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device).

[0154] The above embodiments provide a detailed introduction to the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A high-precision collaborative predictive control method for dual transfer robots, characterized in that: The specific steps include: S1. Build a dual-transfer robot collaborative control system hardware platform, including two transfer robot systems, a binocular camera as a visual servo device, large workpieces to be processed, and an industrial computer equipped with the dual-transfer robot collaborative control system; each transfer robot system includes a six-degree-of-freedom robot, a transfer platform, and an end effector; S2. Use binocular cameras to measure and establish a global coordinate system and local coordinate systems for each component, and construct a coordinate system transformation model to determine the positional relationship mapping between the large workpiece to be processed, the six-degree-of-freedom robot, the end effector, and the transfer platform; S3. Based on the position relationship mapping, the coarse positioning of the transfer platform and the solution of the end effector processing trajectory are performed; S4. Construct a dual-robot cross-coupling collaborative predictive control model to calculate the positioning error of each transfer robot system at the current moment and the collaborative trajectory error between the transfer robot systems, and determine whether the two transfer robot systems have achieved collaboration; then solve the compensation value at the current moment and predict the output position of the end effector at the next moment; S5. Calculate the joint angles of each six-degree-of-freedom robot under the output posture of the end execution at the next moment through the corresponding robot inverse solver; send the joint angles to the corresponding motor drivers to realize the coordinated motion control of the dual transfer robots.

2. A high-precision collaborative predictive control method for dual transfer robots according to claim 1, characterized in that: In step S1: In each transfer robot system: the six-degree-of-freedom robot is fixed on the transfer platform as a whole, and the end effector is installed at the end of the sixth joint of the six-degree-of-freedom robot; The dual-transfer robot collaborative control system integrates functional modules including binocular camera data reading, coordinate system conversion model, tool path planning, dual-robot cross-coupling collaborative predictive control model, robot inverse solver, and control modules for the operation control of six-degree-of-freedom robots, transfer platforms and end effectors.

3. The high-precision collaborative predictive control method for dual transfer robots according to claim 1, characterized in that: Step S2 specifically includes: S21, establish the target feature coordinate system of the large workpiece to be processed {W i }、Six-DOF robot base coordinate system {B i }、End effector coordinate system {E i }、Transfer platform coordinate system {A i }, binocular camera coordinate system {C} and global coordinate system {G}; where i = 1, 2, representing the first and second six-degree-of-freedom robots; S22, through the coordinate system conversion model, obtain the target feature coordinate system {W i }Transformation matrix relative to the global coordinate system {G} Transfer platform coordinate system {A i }Transformation matrix relative to the global coordinate system {G} Transformation matrix of the binocular camera coordinate system {C} relative to the global coordinate system {G} S23, then obtain the coordinate system of the transfer platform {A i }Relative to the target feature coordinate system {W i }'s transformation matrix End effector coordinate system {E i }Transformation matrix relative to the global coordinate system {G} End effector coordinate system {E i }Relative to the target feature coordinate system {W i }'s transformation matrix 4. The high-precision collaborative predictive control method for dual transfer robots according to claim 2, characterized in that: Step S3 specifically includes: S31. Determine the target position information of the large workpiece to be processed based on the transformation matrix of the transfer platform coordinate system relative to the target feature coordinate system; based on this information, the dual transfer robot collaborative control system drives the transfer platform to perform station conversion, and activates the transfer platform support mechanism after reaching the target area to achieve coarse positioning of the transfer platform; S32. Then, according to the position information of the current end effector coordinate system relative to the target feature coordinate system, the machining trajectory is solved by the tool path planner.

5. The high-precision collaborative predictive control method for dual transfer robots according to claim 1, characterized in that: Step S4 specifically includes: S41, at each moment, the actual position of the end effector is monitored and fed back in real time by the binocular camera, and the position information of the end effector at that moment is extracted from the processing trajectory as the ideal position of the end effector, thereby obtaining the positioning error matrix X of the single transfer robot system l , the formula is: Among them, X 1_l =[x i ,y i ,z i ,α i ,β i ,γ i ] T is the pose error matrix composed of the single robot translation error and pose error, For X 1_l Differentiation with respect to time; l is 1 and 2 respectively represent the first and second transfer robot systems; S42. Determine the positioning error of each transfer robot system based on the positioning error model. Then, using one of the dual transfer robots as a master robot and the other as a slave robot, determine the collaborative trajectory error and cross-coupling error between the master and slave robots based on the positioning error of the single transfer robot system. S43. Construct a state space model of the dual-robot collaborative control system. The formula is expressed as: Among them, x(k) and x(k+1) are the collaborative trajectory error states at time k+1 and time k respectively. is the control input, is the controlled output, is the external disturbance, A is the state transfer matrix, B u 、B d are the control input parameter matrix and the external disturbance parameter matrix, C c is the observation matrix; S44. Based on the state space model of the dual-robot collaborative control system and combined with the current state input, predict the controlled output for multiple steps in the future, with the goal of minimizing the cross-coupling error, and solve the optimal control input increment as the compensation value and the output posture of the end effector at the next moment.

6. The high-precision collaborative predictive control method for dual transfer robots according to claim 5, characterized in that: Step S42 is specifically as follows: According to the positioning error model of a single transfer robot system, the positioning error e of the master and slave robots is obtained. r_1 、e r_2 , that is, e r_1 =X1,e r_2 =X2; then define the collaborative trajectory error between the master and slave robots as: ε r_1 =and r_1 -And r_2 ,e r_2 =and r_2 -And r_1 ; Among them, ε r_1 With ε r_2 are the collaborative trajectory errors of the master and slave robots respectively. When ε r_1 = 0 and ε r_2 =0 means that the two robots realize cooperative control; The error matrix of the dual-robot collaborative trajectory is denoted as Δ; the cross-coupling error between the two transfer robot systems is denoted as Γ; the formulas for Δ and Γ are expressed as: C=[e c_1 ,e c_2 ] T =E+μΔ=(I+μT)E; Among them, e c_1 With e c_2 are the cross-coupling errors of the master and slave robots, respectively; μ is a proportional factor used to control the cooperation between the two robots.

7. The high-precision collaborative predictive control method for dual transfer robots according to claim 5, characterized in that: Step S44 is specifically as follows: S441. Calculate the predicted increment of the collaborative trajectory error state variable at time k to time k+1, Δx(k+1|k). The formula is: Δx(k+1∣k)=AΔx(k)+B u Δu(k)+B d Δd(k); in, is the collaborative trajectory error state increment, is the control input increment, Δd(k) is the external disturbance increment; S442, obtain the predicted value of the coordinated trajectory error state at time k for time k+p by recursion, and then substitute it back into the state space model to obtain the predicted value y of the controlled output at time k for time k+p c (k+p|k), ​​the formula is expressed as: S443, perform rolling optimization on the predicted value of the controlled output; when there is a control input increment, optimize y by proportional relationship and superposition principle. c (k+p|k) is simplified to obtain the predicted value of the controlled output at the future moment under the action of the one-step control input increment. The formula is: in, is the predicted value of the controlled output at the future moment when no control input increment is added; a i is the unit step response coefficient, N is the optimization time domain, that is, the number of optimization steps in the future; Δu(k)=u(k)-u(k-1) represents the control input increment at time k relative to time k-1; Then recursively obtain the predicted value of the controlled output at the future moment under the M-step control input increment S444, for Define the performance evaluation index minJ(k); the formula is expressed as: Among them, q i 、r j are the error weighting factor and control constraint factor, which respectively reflect the degree of suppression of the collaborative trajectory error and the constraint strength of the control input increment; w(k+i) is the expected value of the control input increment determined at each time point k for the controlled object at the next P time points; M is the control horizon, that is, the actual number of control adjustment steps and M≤P≤N; remember is a matrix composed of the predicted values ​​of the controlled output in the next P steps under the control input increment, is the future P steps without control input increment The matrix composed of the controlled output prediction values; satisfy: Where Δu M (k) is the M-step control input increment matrix, and the j-th step control input increment is expressed as Δu j (k)=U j T Δu M (k); U j is a row vector whose jth entry is 1 and the rest are 0; according to The performance evaluation index minJ(k) is simplified by the relationship between Among them, w P (k) = [w(k+1)…w(k+P)] T ; Q, R are the error weight matrix and control constraint matrix, Q = diag (q1, ..., q P ), R=diag(r1,…,r M ); At time k, w P (k), All are known, then according to the extreme value necessary condition dJ(k) / dΔu M (k) = 0, find the optimal M-step control input increment matrix that minimizes J(k), denoted as Δu M (k)', the formula is expressed as: That is, the compensation value in the future control time domain of size M is obtained; S445, at time k, Δu M (k)' is the cross-coupling error applied to the lth robot, then the predicted output value y under the action of the optimal control input increment is N1 (k) is: y N1 (k)=y N0 (k)+a1'Δu1(k)'; Among them, y N0 (k) is the controlled output when no control input increment is added; Δu1(k)' is Δu M The first step in (k)' is the control input increment, and a1' is the corresponding unit step response coefficient; The actual output y(k+1) at time k+1 is compared with the predicted output value of the controlled object at time k+1 under the action of the one-step control input increment. By comparison, we can get the prediction error e(k+1) at time k+1; the expression is: Combined with the calculated optimal control input increment matrix, the prediction error is heuristically corrected to obtain the predicted value of the controlled output at time k+1 under the corrected optimal control input increment. for: That is, the output pose of the end effector at time k+1 is obtained; where the correction vector h is an N-dimensional vector consisting of weight coefficients h = [h1 … h N ] T ; The subsequent time steps continue to perform online iterations using the rolling optimization strategy to obtain the output pose for each time step in turn.

Images