Multi-AGV collaborative teaching reproduction navigation method based on UWB positioning and double-layer control, medium and equipment

Abstract

The invention provides a multi-AGV collaborative teaching reproduction navigation method based on UWB positioning and double-layer control, a medium and equipment, and belongs to the technical field of AGV collaborative autonomous navigation. The method comprises the following steps: S1, constructing a global coordinate system; s2, UWB-IMU integrated beacons are installed on the carrying equipment and used for collecting UWB data and IMU data; s3, designing an adaptive Kalman filtering algorithm for fusing the UWB data and the IMU data; s4, entering a teaching phase, guiding a plurality of AGVs to carry out cooperative carrying of equipment, and calculating key point coordinates and motion parameters according to the Kalman filtering algorithm designed in the step S3; and S5, entering a reappearing stage, and tracking a teaching path by adopting MPC double-layer control according to the key point coordinates and the motion parameters obtained in the step S4. The method has the advantages of no need of laying markers, low environmental requirements, high navigation precision and reliability, low price and the like.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of AGV cooperative autonomous navigation, and particularly relates to a multi-AGV cooperative teaching-reproduction navigation method based on UWB positioning and double-layer control, a medium and equipment. BACKGROUND

[0002] In the construction and operation of substations, the handling of heavy loads such as GIS devices (usually weighing ≥ 12 tons) faces the following technical bottlenecks: 1. Single vehicle load limitation: the carrying capacity of a single traditional AGV is generally < 10 tons, which cannot meet the heavy load demand; 2. Trajectory deviation caused by unsynchronized poses during multi-vehicle cooperative handling: during multi-vehicle cooperative handling, synchronization errors accumulate, leading to errors in trajectory tracking; 3. Poor environmental adaptability: single ultra-wideband (UWB) positioning has obvious signal interruption in shielding scenarios, and single inertial navigation (IMU) has long-term cumulative errors. In addition, in the prior art, magnetic tape navigation requires the laying of markers, laser navigation has serious reflection interference in metal environments, and traditional PID control cannot balance trajectory accuracy and control smoothness. Therefore, there is an urgent need for an integrated solution that combines high-precision positioning, multi-vehicle cooperation, and intelligent control. SUMMARY

[0003] The application provides a multi-AGV cooperative teaching-reproduction navigation method based on UWB positioning and double-layer control, a medium and equipment to solve the problems in the prior art.

[0004] To achieve the above-mentioned purposes, the application adopts the following technical solutions:

[0005] In a first aspect, the application provides a multi-AGV cooperative teaching-reproduction navigation method based on UWB positioning and double-layer control, comprising the following steps:

[0006] S1: Construct a global coordinate system;

[0007] S2: Install a UWB-IMU integrated beacon on the handling device for collecting UWB data and IMU data;

[0008] S3: Design an adaptive Kalman filter algorithm for fusing UWB data and IMU data;

[0009] S4: Enter the teaching phase, guide multiple AGVs to cooperatively handle the device, and calculate key point coordinates and motion parameters according to the Kalman filter algorithm designed in step S3;

[0010] S5: Enter the reproduction phase, track the teaching path using MPC double-layer control based on the key point coordinates and motion parameters obtained in step S4.

[0011] Optionally, in step S1, the constructing a global coordinate system is specifically: deploying 4 UWB base stations at the rectangular four corners of the target carrying space to construct an absolute coordinate system covering the whole scene; and establishing a relative coordinate system with the geometric center of the device as the origin, defining the length direction of the device as the X axis and the width direction as the Y axis.

[0012] Optionally, in step S2, the UWB-IMU integrated beacon integrates a UWB positioning unit and an IMU inertial measurement unit, the UWB positioning unit calculates the beacon coordinates by using a double-lateral ranging method, and the IMU inertial measurement unit synchronously acquires IMU data.

[0013] Optionally, in step S3, the adaptive Kalman filtering algorithm includes the following steps:

[0014] S3.1: performing state prediction according to the device state at the last moment:

[0015]

[0016] In the formula, is the prior state estimation; F is the state transition matrix, F T is the transpose matrix of F; x k-1 is the posterior state estimation at the last moment; B is the control input matrix; u k is the control input vector; is the prior estimation covariance; P k-1 is the posterior estimation covariance at the last moment; Q is the process noise covariance matrix;

[0017] S3.2: acquiring the current received UWB data quantity N, and calculating the ranging residual error Z′ k is the filtered measurement vector, is the beacon coordinates, H is the observation matrix, and the UWB data quality Q is evaluated by the following formula: UWB

[0018]

[0019] In the formula, α1 and α2 respectively represent the data quantity stability and the weight of the ranging residual error, N max is the theoretical number of coordinates acquired within one filtering period, d max is the AGV travel distance within one filtering period;

[0020] When Q UWB ≥ 0.7, the UWB data is trusted, and mean filtering is performed on the UWB data;

[0021] When Q UWB < 0.7, it is considered that the UWB data is unreasonable, and the UWB data is fused with the IMU data; ​

[0022] S3.3: dynamically adjust the measurement noise covariance matrix R using the following formula:

[0023] R = R min + (R max - R min ) x (1 - Q UWB );

[0024] wherein R min is the measurement error variance without occlusion, and R max is the measurement error variance when there is severe occlusion.

[0025] Step 3.4: calculate the Kalman gain K k and update the state of the device:

[0026]

[0027] Optionally, in step S4, three AGVs are used for cooperative transportation of the device, and the motion parameters include the heading angle, the left and right wheel speeds, and the motion mode.

[0028] Optionally, in step S5, the MPC double-layer control includes lower-layer real-time pose angle correction and upper-layer MPC optimization control.

[0029] Optionally, the lower-layer real-time pose angle correction is as follows:

[0030] Real-time monitoring of the AGV pose angle deviation φ, and dynamically adjusting the pose angle deviation φ using the differential wheel speed model to ensure that the AGV pose angle is consistent with the device heading angle. The differential wheel speed model is as follows:

[0031]

[0032] wherein ω e is the differential wheel angular velocity; Δt is the lower-layer control period; v L and v R are the left and right wheel speeds, respectively; v is the AGV centerline speed; and L is the wheelbase.

[0033] Optionally, the upper-layer MPC optimization control is to optimize the speeds of the three vehicles through MPC to minimize the teaching trajectory tracking error and control input fluctuation, and is as follows:

[0034] Set the device reference trajectory heading angle as θ ref , and the device state vector X d = [x d , y d , θ d ] T , wherein (x d , y d ) are the key point coordinates, and θ dLet u be the heading angle, and let u = [v1, v2, v3] be the control input vector. T v1, v2, and v3 represent the speeds of the three vehicles, respectively. The state of the equipment at the next moment is:

[0035]

[0036] In the formula, k and k+1 represent the current time and the next time, respectively; r is the distance between the AGV and the center of the equipment; α is the angle between the AGV speed direction and the line r; ΔT is the upper-level MPC control cycle; and the subscript i represents the AGV sequence number.

[0037] The state error term and the control input change term are constructed as follows:

[0038]

[0039] In the formula, X ref Q is the device state vector collected during the teaching phase. j Let R be the state error weight matrix. J The weight matrix is ​​the control input change matrix, and u(k) is the control input vector at time k;

[0040] Based on the state error term and the control input change term, the cost function J in the prediction time domain N is constructed as follows:

[0041]

[0042] The cost function is solved using an iterative search algorithm to find the optimal control input; the control input vector u in the prediction time domain is defined. N =[u0,u1,…,u 3N-1 ] T , where u 3N-3 ,u 3N-2 ,u 3N-1 This represents the prediction of the speeds of the three vehicles at the Nth time point in the time domain; M iterations are performed, and the control input vector for the mth iteration is... The i-th component is increased or decreased by a perturbation step size δ, while the other components remain unchanged. The j-th component is:

[0043]

[0044] Then calculate the cost function values ​​after forward and reverse perturbations. and Compare the current optimal cost function value J best and The values ​​of these three factors are considered, and the control input vector that minimizes the cost function value is taken as the new optimal solution. The new optimal control input vector is then calculated. And the optimal control input vector of the previous iteration the standard deviation of u N :

[0045]

[0046] If Δu N ≤ ∈ u , ∈ u denotes the control input limit, iteration convergence, and the first three elements of u are taken as the optimal control input; if Δu> ∈ u , the operation is repeated until the maximum iteration number M, and the first three elements of u are taken as the optimal control input.

[0047] In a second aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program enables a computer to execute the multi-AGV collaborative demonstration reproduction navigation method based on UWB positioning and double-layer control as described in the first aspect.

[0048] In a third aspect, the present application provides an electronic device, which comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the multi-AGV collaborative demonstration reproduction navigation method based on UWB positioning and double-layer control as described in the first aspect is realized.

[0049] The present application has the advantages that the present application solves the problems of limited carrying capacity, poor flexibility, and low efficiency of a single differential AGV in carrying heavy long equipment in a limited space, and the problems of more than three-point support of a "two master two slave" system and low safety factor due to the lack of suspension of the vehicle body. The present application constructs a UWB global coordinate system and a device relative coordinate system, fuses UWB and IMU data using dynamic weight adaptive Kalman filtering, effectively improves the positioning accuracy, solves the problems of obvious signal interruption of single UWB positioning in shielding scenes and long-time cumulative error of single inertial navigation, realizes high-precision collaborative carrying of multiple AGVs through double-layer MPC control, solves the problem of trajectory deviation caused by different poses of multiple vehicles during collaborative carrying, does not need to lay markers, reduces the requirements on the environment, and is suitable for the collaborative carrying of heavy loads such as GIS equipment and reactors in substations, overcomes the defects of limited carrying capacity of traditional single AGV, has more advantages than magnetic tape navigation and laser navigation, and solves the problem that traditional PID control is difficult to balance trajectory accuracy and control smoothness. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 is a work scene layout diagram of the present application;

[0051] Figure 2 is a global and local kinematics model diagram;

[0052] Figure 3 The figure is a fusion principle flow chart of UWB data and IMU data.

[0053] Figure 4 The figure is a flow chart of the demonstration stage.

[0054] Figure 5 The figure is a flow chart of the reproduction stage. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application.

[0056] In an embodiment, the present application provides a multi-AGV collaborative demonstration and reproduction navigation method based on UWB positioning and double-layer control. The UWB navigation technology converts the distance between the beacon and the multi-base station by calculating the wave propagation time of the electromagnetic wave transmission, obtains the absolute coordinates of the beacon in the two-dimensional plane, and has the advantages of no cumulative error, strong anti-interference ability, low power consumption, low cost, etc. The navigation process is divided into two steps: first, after the base station is built and the indoor two-dimensional coordinate map is established, the AGV car is remotely controlled to complete a demonstration movement, and a series of track coordinates and corresponding control commands of the car are recorded; in the reproduction stage, the multi-AGV is controlled by the multi-target MPC control with the minimum distance and heading angle deviation, the movement trajectory is reproduced, and the trajectory tracking is realized. Since the demonstration method is adopted, the optimal route in a wide range can be obtained to avoid obstacles, and the dynamic obstacles can also be detected in real time by the ultrasonic sensor on the vehicle and stopped urgently. The method has the advantages of no need to lay markers, low environmental requirements, high navigation accuracy and reliability, low price, etc.

[0057] The method realizes high-precision collaborative carrying of heavy equipment through the following steps:

[0058] S1: Construct a global coordinate system.

[0059] As shown in Figure 1 , four UWB base stations are deployed at the rectangular corners of the target carrying space to construct an absolute coordinate system covering the whole scene; a relative coordinate system is established with the geometric center O of the equipment as the origin, and the length direction of the equipment is defined as the X-axis (the forward direction is positive), and the width direction is defined as the Y-axis (the left direction is positive).

[0060] S2: Sensor installation and information acquisition.

[0061] Further referring to Figure 1A UWB-IMU integrated beacon is rigidly installed at the center of the top of the device, and real-time pose data is transmitted to the AGV controller through a 2.4G wireless module. The beacon integrates a UWB positioning unit and an IMU inertial measurement unit. The positioning adopts a double ranging method to calculate the beacon coordinates (i.e. UWB data), and at the same time, IMU data is obtained.

[0062] S3: An adaptive Kalman filtering algorithm is designed to fuse UWB and IMU data, as shown in Figure 3 , which includes the following steps:

[0063] (1) State prediction stage:

[0064] According to the state X k-1 of the handling device at the last moment, the state is predicted:

[0065]

[0066] Among them: is the prior state estimate; F is the state transition matrix; X k-1 is the posterior state estimate at the last moment; B is the control input matrix; u k is the control input vector; is the prior estimate covariance; P k-1 is the posterior estimate covariance at the last moment; Q is the process noise covariance matrix.

[0067] (2) UWB data quality evaluation:

[0068] Get the current received UWB data N, calculate the ranging residual error Evaluate the quality of UWB data by the following formula:

[0069]

[0070] Among them, α1, α2 represent the weight of data stability and ranging residual error respectively, take α1=0.6, α2=0.4; N max is the number of theoretical coordinates obtained in a filtering period; d max is the distance traveled by the trolley in a filtering period.

[0071] When the data quality Q UWB is greater than 0.7, the UWB coordinate data is trusted at this moment, and the mean filtering is performed on the last 5 UWB observation values, and the jump value deviating from the mean value by more than 3 times the standard deviation is removed.

[0072] When the data quality Q UWB is less than 0.7, it is considered that the UWB data is unreasonable at this moment, and the IMU data needs to be fused.

[0073] (3) The measurement noise covariance matrix R is dynamically adjusted using the following formula:

[0074] R = R min +(R max -R min )×(1-Q UWB )

[0075] Where: R min For the variance of unobstructed measurement error, r max The variance of measurement error under severe occlusion.

[0076] (4) Calculate the Kalman gain K during the update phase. k And update the status:

[0077]

[0078] Where: H is the observation matrix, X k For the posterior estimation of the state, Z′ k This is the preprocessed measurement vector.

[0079] S4: As Figure 4 As shown, during the teaching data acquisition phase, the operator guides multiple AGVs (3 in this embodiment) to collaboratively transport materials via remote control. Based on the Kalman filter algorithm in S3, the coordinates (x, y, y) of the key points after Kalman filtering are calculated with a period of 500ms. d ,y d ) and motion parameters (heading angle θ) d (Left and right wheel speeds and motion mode); motion mode flag (0 = translation mode, 1 = center rotation mode), data is stored in the buffer of the main control unit, and power failure protection is supported.

[0080] S5: In the reproduction phase, MPC two-layer control is used to implement teaching path tracking. (Refer to...) Figure 2 , Figure 5 As shown, the MPC two-layer control reproduction strategy includes the following steps:

[0081] (1) Real-time correction of lower layer pose angle:

[0082] The AGV's position angle deviation φ is monitored in real time, and dynamically adjusted using a differential wheel speed model to ensure that the AGV's position angle remains consistent with the equipment's heading angle. The differential wheel model is as follows:

[0083]

[0084] Where: ω e V is the differential wheel angular velocity; Δt is the lower-level control cycle; v L v R is the left and right wheel speeds; v is the AGV center linear speed; L is the wheel track.

[0085] Upper MPC optimization control:

[0086] Multi-vehicle tracking trajectory and equipment heading angle have been obtained by teaching reproduction record. Through MPC optimization of three vehicle speeds, the teaching trajectory tracking error and control input fluctuation are minimized. Set the equipment reference trajectory heading angle as θ ref In each control period k, the equipment state vector X d = [x d , y d , θ d ] T , the control input vector u = [v1, v2, v3] T , v1, v2, v3 represent the speeds of three vehicles respectively. The next time state is:

[0087]

[0088] Where: r is the distance between AGV and equipment center; α is the angle between AGV speed direction and the line where r is located, ΔT is the upper MPC control period.

[0089] Construct state error term and control input change term:

[0090]

[0091] Where: Q j is the state error weight matrix; R J is the control input change weight matrix.

[0092] Construct the cost function J in the prediction horizon N:

[0093]

[0094] Where: X ref (k) is the equipment state vector collected during the teaching process.

[0095] Solve the optimal control sequence by iterative search algorithm, define the control input vector u N = [u0, u1, …, u 3N-1 ] T , where N is the prediction horizon, u 3N-3 , u 3N-2 , u 3N-1 represent the speeds of three vehicles at the Nth time in the prediction horizon. Perform M iterations, in each iteration m (m = 1, 2, …, M), respectively, add or subtract a disturbance step size δ to each component of the control input vector , and keep the rest unchanged, that is:

[0096]

[0097] Calculate the cost function value after forward and backward perturbation and Compare the optimal cost function J best with the size of the three, take the control input vector u (m) that makes the cost function minimum as the new optimal solution. Establish the control input change constraint condition:

[0098] Δu=||u (m) -u (m-1) ||2

[0099] If Δu≤∈ u (control input amplitude limiting), the iteration converges; if Δu>∈ u , repeat the above steps until convergence or reach the maximum iteration number M. Take the first three elements of u (m) as the optimal control input u opt at the next moment:

[0100] u opt =[u (M) [0],u (M) [1],u (M) [2]] T

[0101] The three vehicles cooperatively carry out the above control process until reaching the destination. After the equipment is in place, the worker issues a reset command through the remote controller, and the three vehicles automatically return to the initial position, waiting for the next carrying to start.

[0102] In another embodiment, the present application provides a computer readable storage medium, which stores a computer program, and the computer program causes a computer to execute the multi-AGV cooperative demonstration reproduction navigation method based on UWB positioning and double-layer control of the foregoing embodiment.

[0103] In another embodiment, the present application provides an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the multi-AGV cooperative demonstration reproduction navigation method based on UWB positioning and double-layer control of the foregoing embodiment when executing the computer program.

[0104] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium which can contain or store programs for use by or in connection with an instruction execution system, apparatus or device. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any suitable combination of the above. More specific examples of computer storage medium can include one or more wires, portable computer disks, hard drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), fiber optics, compact disk read-only memory (CDROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0105] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present application can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software manner depends on the specific application and design constraints of the technical solutions. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0106] The above is only the preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solution falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that, for ordinary skilled persons in the art, some improvements and refinements without departing from the principles of the present application shall be considered within the protection scope of the present application.

Claims

1. A method for multi-AGV collaborative teaching and reproduction navigation based on UWB positioning and double-layer control, characterized in that, Comprising the following steps: S1: constructing a global coordinate system; S2: installing a UWB-IMU integrated beacon on the handling equipment for collecting UWB data and IMU data; S3: designing an adaptive Kalman filtering algorithm for fusing UWB data and IMU data; S4: entering the teaching stage, guiding multiple AGVs to cooperatively handle the equipment, and calculating key point coordinates and motion parameters according to the Kalman filtering algorithm designed in step S3; S5: entering the reproduction stage, tracking the teaching path using MPC double-layer control according to the key point coordinates and motion parameters obtained in step S4. 2.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method according to claim 1, characterized in that: In step S1, the construction of the global coordinate system specifically comprises: deploying four UWB base stations at the corners of the target handling space to construct an absolute coordinate system covering the entire scene; and establishing a relative coordinate system with the geometric center of the equipment as the origin, defining the length direction of the equipment as the X-axis and the width direction as the Y-axis. 3.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method according to claim 1, characterized in that: In step S2, the UWB-IMU integrated beacon integrates a UWB positioning unit and an IMU inertial measurement unit, the UWB positioning unit calculates the beacon coordinates using the double-side ranging method, and the IMU inertial measurement unit synchronously obtains IMU data. 4.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method of claim 3, wherein: In step S3, the adaptive Kalman filtering algorithm comprises the following steps: S3.1: predicting the state according to the equipment state at the previous time; wherein is the prior state estimate; F is the state transition matrix, F T is the transpose of F; X k-1 is the prior state estimate; B is the control input matrix; u k is the control input vector; is the prior estimate covariance; P k-1 is the prior estimate covariance; Q is the process noise covariance matrix; S3.2: Obtain the current received UWB data amount N, calculate the ranging residual Z' k For the filtered measurement vector, for the beacon coordinates, H is the observation matrix, and the UWB data quality Q is evaluated by the following formula UWB : In the formula, α1, α2 respectively represent the weight of data quantity stability and ranging residual error, N max is the number of theoretically obtained coordinates in a filtering period, d max is the AGV travel distance in a filtering period; When Q UWB ≥ 0.7, trust the UWB data and perform mean filtering on the UWB data; When Q UWB When UWB data is considered unreasonable, fuse UWB data with IMU data; S3.3: dynamically adjusting the measurement noise covariance matrix R using the following formula: R=R min +(R max -R min )×(1-Q UWB ); where R min is the unobstructed measurement error variance, R max is the measurement error variance when heavily obstructed. Step 3.4: Calculate Kalman gain K k and update the state of the device: 5.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method according to claim 1, characterized in that: In step S4, three AGVs are used to cooperatively handle the equipment, and the motion parameters include the heading angle, left and right wheel speeds, and motion mode. 6.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method according to claim 5, characterized in that: In step S5, the MPC double-layer control includes real-time deviation correction of the pose angle in the lower layer and MPC optimization control in the upper layer. 7.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method according to claim 6, characterized in that: The real-time deviation correction of the pose angle in the lower layer is specifically as follows: Real-time monitoring of the AGV pose angle deviation φ, dynamic adjustment of the pose angle deviation φ using the differential wheel speed model to ensure that the AGV pose angle is consistent with the equipment heading angle, and the differential wheel speed model is as follows: In the formula, ω e is the differential wheel angular velocity; Δt is the lower layer control period; v L , v R are the left and right wheel speeds, respectively; v is the AGV centerline speed; and L is the wheel track. 8.The UWB positioning and double-layer control based multi-AGV cooperative teaching and reproduction navigation method according to claim 6, characterized in that: The MPC optimization control in the upper layer is to optimize the speeds of the three vehicles through MPC to minimize the teaching trajectory tracking error and control input fluctuation, which is specifically as follows: Set the device reference trajectory heading angle as θ ref , the device state vector X d = [x d , y d , θ d ] T , (x d , y d ) is the key point coordinate, θ d is the heading angle, the control input vector u = [v1, v2, v3] T , v1, v2, v3 respectively represent three vehicle speeds, and the next state of the device is: Where k and k+1 represent the current time and the next time respectively, r is the distance between the AGV and the equipment center, α is the included angle between the AGV speed direction and the straight line along which r lies, ΔT is the upper MPC control period, and subscript i represents the AGV serial number; The state error term and the control input change term are constructed as follows: where X ref is the device state vector collected in the demonstration phase, Q j is the state error weight matrix, R J is the control input change weight matrix, and u(k) is the control input vector at time k. According to the state error term and the control input change term, the cost function J in the prediction time domain N is constructed as follows: The cost function is solved using an iterative search algorithm to find the optimal control input; the control input vector u in the prediction time domain is defined. N =[u0,u1,…,u 3N-1 ] T , where u 3N-3 ,u 3N-2 ,u 3N-1 This represents the prediction of the speeds of the three vehicles at the Nth time point in the time domain; M iterations are performed, and the control input vector for the mth iteration is... The i-th component is increased or decreased by a perturbation step size δ, while the other components remain unchanged. The j-th component is: Further, the cost function values after the forward and reverse perturbations are calculated And The current best cost function value J is compared best With The sizes of the three, and the control input vector that makes the cost function value minimum is taken as the new optimal solution, and the new optimal control input vector And the standard deviation Δu of the optimal control input vector of the last iteration N :​ if Δu N ≤ ∈ u , ∈ u denotes the control input clipping, the iteration converges, take the first three elements of as the optimal control input; if Δu> ∈ u , repeat the operation until the maximum iteration number M, take the first three elements of as the optimal control input.

9. A computer readable storage medium storing a computer program, characterized in that, The computer program enables the computer to execute the multi-AGV cooperative teaching and reproduction navigation method based on UWB positioning and double-layer control as claimed in any one of claims 1-8.

10. An electronic device, comprising: Comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the multi-AGV cooperative teaching and reproduction navigation method based on UWB positioning and double-layer control as claimed in any one of claims 1-8 when executing the computer program.

Images