Mobile parallel robot with high load pointing positioning function and control method

Abstract

The application discloses a mobile parallel robot with high-load pointing positioning function and a control method, and the robot comprises a three-wheeled vehicle body, three universal wheels and a vehicle leveling device are installed on the outer wall of the vehicle body, a space posture adjusting device is installed on the upper surface of the vehicle body, a force sensor top plate is installed on the top of the space posture adjusting device, and a force sensor and a moving platform are installed on the upper surface of the force sensor top plate. The control method comprises the following steps: establishing an institution and a universal wheel inverse solution; solving a second derivative to obtain acceleration information; obtaining human-machine interaction force signals from the force sensor and carrying out denoising processing; extracting state characteristics and initializing adaptive parameters; combining layered adaptive admittance control of state characteristics; designing a gradient optimization evaluation function; and controlling and driving motors. The method effectively overcomes fluctuations and disturbances under high-load working conditions, and realizes high-precision pointing positioning and smooth motion control of the robot.

Description

Technical Field

[0001] This invention belongs to the field of robotics technology, specifically relating to a mobile parallel robot with high-load pointing and positioning capabilities and its control method. Background Technology

[0002] Currently, with the increasing demands for attitude adjustment and assembly in the manufacturing, maintenance, and repair of large components such as vehicle chassis, battery packs, and wings / sections, mobile robots are gradually shifting from simple material handling to high-precision mobile operations. Existing mainstream solutions mostly consist of a combination of a "mobile platform + serial robotic arm," but these have shortcomings in terms of rigidity, center of gravity height, and space occupancy.

[0003] Currently, some mobile robots cannot simultaneously possess high load capacity, high-precision positioning, and compliant human-robot collaboration capabilities. For example, the AMR devices described in Chinese patents CN217776987U and CN113858178A, while having a large end-effector range of motion, suffer from poor rigidity in their serial structure, making it difficult to meet the demands of heavy-load assembly. Similarly, the mobile robot described in Chinese patent CN104802151A lacks force sensing capabilities, rendering it unable to effectively perceive external stimuli and thus lacking human-robot interaction capabilities. Summary of the Invention

[0004] This invention is proposed to solve the problems existing in the prior art. Its purpose is to provide a mobile parallel robot and control method with high load-bearing capacity, high-precision positioning and compliant human-machine collaboration capabilities, and high-load pointing and positioning function.

[0005] The technical solution of the present invention is: a mobile parallel robot with high load pointing and positioning function, including a three-wheeled vehicle body, three universal wheels installed on the outer wall of the vehicle body, three vehicle body leveling devices installed around the vehicle body, a spatial posture adjustment device installed on the upper surface of the vehicle body, a force sensor top plate installed on the top of the spatial posture adjustment device, a force sensor installed on the upper surface of the force sensor top plate, and a moving platform for spatial position adjustment output installed on the top of the force sensor.

[0006] The spatial attitude adjustment device includes three branches with the same structure: a first branch, a second branch, and a third branch. The first branch, the second branch, and the third branch are arranged in a staggered 60° arrangement between each other.

[0007] Each of the three branches includes a branch connector fixed to the upper surface of the vehicle body, and the upper end of the branch connector is connected to the fourth rotating ball joint and the second rotating ball joint.

[0008] One end of the motor mounting plate is connected to the motor, and the other end is fixed to the sleeve. The tail of the sleeve is connected to the fourth rotating ball joint, and a sliding joint for output is installed in the sleeve.

[0009] The movable joint is connected to the fixed end of the movable joint, the fixed end of the movable joint is connected to the RRR lever by the third rotating ball joint, the RRR lever is connected to the RS link by the first rotating ball joint, and the RRR lever is connected to the second rotating ball joint.

[0010] The other end of the RS link is connected to a ball joint, and the other end of the ball joint is connected to the three vertices of the top plate of the force sensor.

[0011] Furthermore, the vehicle body is equipped with a battery module, a robot controller, and a universal wheel motor.

[0012] The control method for a mobile parallel robot with high-load pointing and positioning function provided by the present invention includes the following steps:

[0013] (A) Establish the inverse kinematics of the spatial attitude adjustment device, establish the inverse kinematics of the omnidirectional wheel, and introduce the load compensation coefficient. The calculation formula is:

[0014] ,

[0015] In the formula, The current load weight. This is the rated load for the space attitude adjustment device;

[0016] Calculate the position information of the end effector platform at each moment. Solve for the first derivative to obtain the velocity amplitude, and then smooth the velocity information. In the formula The value is 0.05;

[0017] (B) Continue solving for the second derivative to obtain the acceleration amplitude. When there is a disturbance in the acceleration, an acceleration correction factor is introduced for correction.

[0018] ,

[0019] In the formula , For the corrected acceleration, For high load inertial force, The weight of the robot's end effector;

[0020] (C) The human-computer interaction force signal is obtained from the force sensor, and noise is denoised using the moving average filtering method to obtain the human-computer interaction force amplitude; the first derivative of the interaction force amplitude is calculated to obtain the rate of change of the interaction force; the noise reduction formula is:

[0021] ,

[0022] in For the interaction force after noise reduction, Set the sliding window size to 5-10 sampling points. For the sampling period, extract the amplitude of the human-computer interaction force: Perform the first derivative of the interaction force amplitude to obtain the rate of change of the interaction force:

[0023] ,

[0024] Simultaneously, record the fluctuation range of the rate of change of the interaction force:

[0025] This is used for dynamic adjustment of the subsequent reward function;

[0026] (D) State feature extraction and adaptive parameter initialization:

[0027] The amplitude of the filtered velocity information from step (A), the amplitude of the corrected acceleration information from step (B), and the amplitude and rate of change of the human-computer interaction force after denoising from step (C) are extracted as the real-time operating state features of the system. In order to achieve dynamic disturbance rejection under high load, the system adopts an adaptive parameter optimization framework based on reward gradient, initializes the admittance reference parameters of the spatial attitude adjustment device and the omnidirectional wheel, namely: the initial values ​​of the damping matrix and stiffness matrix, and sets the learning rate used to control the step size of parameter updates. , The value is set between 0.9 and 0.95, and is used as the basis for real-time dynamic adjustment of the admittance parameters.

[0028] (E) Hierarchical adaptive admittance control combining state characteristics:

[0029] For a spatial attitude adjustment device and omnidirectional wheels, a hierarchical adaptive admittance control structure based on state feedback is designed. First, according to the system's degree of freedom configuration, the end motion extracted in steps (A) and (B) is kinematically decoupled: Z-axis translation, X-axis rotation, and Y-axis rotation components are extracted to construct the motion state vector of the spatial attitude adjustment device. In the formula This is the position / attitude vector of the end effector platform, corresponding to the real-time position / angle of the three degrees of freedom: Z-axis translation, X-axis rotation, and Y-axis rotation. The first derivative corresponds to the velocity / angular velocity of the three degrees of freedom; The second derivative corresponds to the acceleration / angular acceleration of these three degrees of freedom; Corresponding to the interaction force / torque components in these three degrees of freedom;

[0030] Extract the X-axis movement, Y-axis movement, and Z-axis rotation components of the end effector platform to construct the motion state vector of the omnidirectional wheel base. In the formula This is the position / attitude vector of the omnidirectional wheel, corresponding to the real-time position / angle of the three degrees of freedom: X-axis movement, Y-axis movement, and Z-axis rotation. The first derivative corresponds to the velocity / angular velocity of the three degrees of freedom; The second derivative corresponds to the acceleration / angular acceleration of these three degrees of freedom; Corresponding to the interaction force / torque components in these three degrees of freedom;

[0031] Admittance models were then established respectively:

[0032] (a) Establish an admittance model for the Z-axis movement, X-axis rotation, and Y-axis rotation of the spatial attitude adjustment device:

[0033] ,in The mass matrix of the spatial attitude adjustment device is 3×3, which is determined by the structural parameters of the mechanism and the high-load mass. The corresponding interaction force components extracted in step (C); The time-varying damping matrix and time-varying stiffness matrix are 3×3, respectively. To adapt to high load fluctuation conditions, the adaptive optimization framework constructed in step (D) is used to dynamically update the admittance parameters in real time along the gradient direction of the evaluation function. The update law is:

[0034] ,

[0035] In the formula, The learning step size for the spatial attitude adjustment device was optimized and determined through high-load operating condition experiments. The evaluation function for the space attitude adjustment device is shown in step (F); during the system initialization phase, the damping matrix and stiffness matrix are preset. The initial values ​​are obtained; within each control cycle of equipment operation, the time-varying damping matrix and time-varying stiffness matrix at the current moment can be obtained. ;

[0036] (b) Establish independent admittance models for the X-axis movement, Y-axis movement, and Z-axis rotation of the omnidirectional wheel:

[0037] ,

[0038] In the formula The mass matrix of the omnidirectional wheel is 3×3; These are the time-varying damping matrix and the time-varying stiffness matrix, both 3×3; the load disturbance of the omnidirectional wheel is dynamically updated in real time by an independent adaptive optimization algorithm, with the update law being:

[0039] ,

[0040] In the formula, The learning step length of the omnidirectional wheel is set independently from the learning rate of the spatial attitude adjustment device and is determined through high-load movement experiments. The evaluation function for the omnidirectional wheel is shown in step (F); during the system initialization phase, the damping matrix and stiffness matrix are preset. The initial values ​​are obtained; within each control cycle of equipment operation, the time-varying damping matrix and time-varying stiffness matrix at the current moment can be obtained. ;

[0041] (F) Gradient optimization evaluation function design: For the spatial attitude adjustment device and the omnidirectional wheel, a system evaluation function for calculating the gradient is designed:

[0042] (a) Evaluation function for spatial attitude adjustment device, focusing on the optimization of attitude error and Z-axis displacement error, the specific formula is as follows:

[0043] ,

[0044] In the formula, The weighting coefficients were determined through repeated optimization experiments using high-load pointing positioning. In practical applications, there is a slight emphasis; These are the actual X-axis rotation angle, actual Y-axis rotation angle, and actual Z-axis movement height of the end-effector of the spatial attitude adjustment device. These represent the desired X-axis rotation angle, desired Y-axis rotation angle, and desired Z-axis movement height of the end effector platform of the spatial attitude adjustment device. This is the sensitivity coefficient, used to adjust the sensitivity of the evaluation function to errors. For the attitude error of the radius, a value of 50 is used. For displacement errors at the millimeter level, a value of 0.03 is used;

[0045] (b) The evaluation function for the omnidirectional wheel focuses on optimizing the position error and heading angle error. The specific formula is as follows:

[0046] ,

[0047] In the formula, The weighting coefficients were determined through repeated optimization experiments under high load. In practical applications, there is a slight emphasis; These represent the actual X-axis movement position, actual Y-axis movement position, and actual Z-axis rotation angle of the omnidirectional wheel, respectively. These represent the desired X-axis movement position, desired Y-axis movement position, and desired Z-axis rotation angle of the omnidirectional wheel, respectively. The sensitivity coefficient is used to adjust the sensitivity of the evaluation function to errors. For the attitude error of the radius, a value of 50 is used. For displacement errors at the millimeter level, a value of 0.03 is used;

[0048] (G) Motor control and drive

[0049] The spatial attitude adjustment device and the omnidirectional wheel adopt the terminal sliding mode control law to ensure that the joint angle tracking error converges within a finite time. The specific formula is as follows:

[0050] ,

[0051] In the formula, To account for the angle tracking error between the spatial attitude adjustment device and the omnidirectional wheel, For the desired angle of the spatial attitude adjustment device and the omnidirectional wheels, The actual angle acquired by the motor encoder; The coefficients are power coefficients, determined through experimental optimization. To control the gain, it was optimized and determined through high-load operating condition experiments; This is a sign function used to implement disturbance suppression in sliding mode control;

[0052] The formula for converting control torque commands into motor current commands is:

[0053] ,

[0054] In the formula, The torque constant of the motor is... The current command is sent to the motor driver to drive the motor of the spatial adjustment device and the universal wheel to rotate, which is the reduction ratio of the reducer.

[0055] Furthermore, in step (A), the inverse solution of the spatial attitude adjustment device is established, specifically as follows:

[0056] Taking the center of the three branches as the origin One branch parallel to the spatial attitude adjustment device is The axis is perpendicular to the branch and parallel to the upper surface of the vehicle body. The axis is perpendicular to the upper surface of the vehicle body. Establish a static coordinate system based on the axes. ;

[0057] The origin is taken as the center of the ball joints of the three branches connected to the moving platform, namely the first, second, and third branches. Establish a moving coordinate system The axes are all parallel to the static coordinate system. ;

[0058] Let the coordinates of the first rotating spherical pair be... These correspond to the first branch, the second branch, and the third branch, respectively; let the coordinates of the second rotating spherical joint be denoted as . Let be the coordinates of the ball joints of the three branches relative to the moving platform, corresponding to the first, second, and third branches respectively. ;

[0059] The offset vector of the moving coordinate system relative to the static coordinate system is ;

[0060] The position vector of the ball joint with three branches in the static coordinate system is: In the formula, For the rotation matrix of the moving platform.

[0061] The vector from the first rotating ball joint to the ball hinge is The vector from the second revolute joint to the first revolute joint is... The closed-loop vector is: ;

[0062] The motion vector of the sliding joint is The ratio of the long end to the short end of the RRR lever is... The distance from the second spherical joint to the first spherical joint is the long end; the distance from the second spherical joint to the third spherical joint is the short end, with a length of... Then the distance traveled by the sliding joint can be obtained as:

[0063] ;

[0064] The distance that the sliding pair needs to move can be converted into the angle that the motor needs to rotate.

[0065] The beneficial effects of this invention are as follows:

[0066] This invention discloses a mobile parallel robot with high-load human-machine collaboration capabilities. Its three-wheeled body has three degrees of freedom: X-axis translation, Y-axis translation, and Z-axis rotation. The spatial attitude adjustment device also has three degrees of freedom: Z-axis translation, X-axis rotation, and Y-axis rotation. This invention effectively eliminates mechanical jitter and abrupt changes under high-load conditions by introducing a load compensation coefficient and signal smoothing and denoising. It employs a hierarchical adaptive admittance control and reward gradient optimization framework, combined with a differentiated weight evaluation function, significantly enhancing the system's compliance and multi-dimensional collaborative positioning accuracy under complex disturbances.

[0067] This invention combines the advantages of high load capacity, high positioning accuracy, and convenient human-machine collaboration, and can be applied to scenarios such as assembly with high-precision pose fine adjustment. Compared with the prior art, this invention meets the needs of large component pose adjustment and assembly, takes into account both high-precision positioning and smooth human-machine interaction, and has high-load human-machine collaboration function. Attached Figure Description

[0068] Figure 1This is a schematic diagram of the structure of a mobile parallel robot with high-load pointing and positioning function provided in an embodiment of the present invention;

[0069] Figure 2 This is a schematic diagram of the internal structure of the vehicle body provided in an embodiment of the present invention;

[0070] Figure 3 This is a schematic diagram of the spatial posture adjustment device provided in an embodiment of the present invention;

[0071] Figure 4 This is a schematic diagram of the branch of the spatial posture adjustment device provided in an embodiment of the present invention;

[0072] Figure 5 This is a flowchart of a control method for a mobile parallel robot with high-load pointing and positioning function provided in an embodiment of the present invention.

[0073] in:

[0074] 1. Vehicle body; 2. Casters; 3. Vehicle body leveling device; 4. Camera; 5. LiDAR; 6. Spatial attitude adjustment device; 7. Force sensor top plate; 8. Force sensor; 9. Moving platform; 10. Battery module; 11. Robot controller; 12. Caster wheel motor; 13. First branch; 14. Second branch; 15. Third branch; 16. Ball joint; 17. RS link; 18. First rotating ball joint; 20. RRR lever; 21. Second rotating ball joint; 22. Third rotating ball joint; 23. Fixed end of sliding joint; 24. Sliding joint; 25. Sleeve; 26. Fourth rotating ball joint; 27. Motor mounting plate; 28. Motor; 29. ​​Pulley; 30. Branch connection seat. Detailed Implementation

[0075] The present invention will now be described in detail with reference to the accompanying drawings and embodiments:

[0076] like Figures 1 to 5 As shown, a mobile parallel robot with high-load pointing and positioning function includes a three-wheeled vehicle body 1, three universal wheels 2 installed on the outer wall of the vehicle body 1, three vehicle body leveling devices 3 installed around the vehicle body 1, a spatial attitude adjustment device 6 installed on the upper surface of the vehicle body 1, a force sensor top plate 7 installed on the top of the spatial attitude adjustment device 6, a force sensor 8 installed on the upper surface of the force sensor top plate 7, and a moving platform 9 for spatial position adjustment output installed on the top of the force sensor 8.

[0077] The spatial attitude adjustment device 6 includes three branches with the same structure: a first branch 13, a second branch 14, and a third branch 15. The first branch 13, the second branch 14, and the third branch 15 are arranged in a staggered manner at 60° between each other.

[0078] Each of the three branches includes a branch connection seat 30 fixed to the upper surface of the vehicle body 1. The upper end of the branch connection seat 30 is connected to the fourth rotating ball joint 26 and the second rotating ball joint 21.

[0079] One end of the motor mounting plate 27 is connected to the motor 28, and the other end is fixed to the sleeve 25. The tail of the sleeve 25 is connected to the fourth rotating ball joint 26. The sleeve 25 is equipped with a sliding joint 24 for output.

[0080] The movable joint 24 is connected to the fixed end 23 of the movable joint. The fixed end 23 of the movable joint is connected to the RRR lever 20 by the third rotating ball joint 22. The RRR lever 20 is connected to the RS link 17 by the first rotating ball joint 18. The RRR lever 20 is connected to the second rotating ball joint 21.

[0081] The other end of RS link 17 is connected to ball joint 16, and the other end of ball joint 16 is connected to the three vertices of the force sensor top plate 7.

[0082] Furthermore, the vehicle body 1 is equipped with a battery module 10, a robot controller 11, and a universal wheel motor 12.

[0083] Specifically, three omnidirectional wheels 2 are evenly arranged around the vehicle body 1 at a 120° angle, and three vehicle body leveling devices 3 are evenly arranged around the vehicle body 1 at a 120° angle. The omnidirectional wheels 2 and the vehicle body leveling devices 3 are evenly arranged in an alternating pattern.

[0084] As one implementation method, the robot can use, but is not limited to, LiDAR 5, camera 4, etc. for navigation.

[0085] As one implementation method, when the robot moves to the designated position, the robot can be fixed in place using the vehicle leveling device 3, which can prevent the vehicle from shaking when the robot is being dragged in a human-robot collaborative manner.

[0086] As one implementation method, the LiDAR 5 scans the environment and creates a map, while the robot uses SLAM navigation. After reaching the designated location, the camera 4 scans the artificially placed QR codes / coded points in the environment to identify the specific location of the three-wheeled vehicle 1, and then performs fine-tuning of the three-wheeled vehicle 1 to avoid errors caused by the low accuracy of SLAM mapping and navigation.

[0087] In one implementation, the vehicle leveling device 3 may include, but is not limited to, the use of electric cylinders or hydraulic cylinders. After the three-wheeled vehicle body 1 reaches the designated position and completes fine-tuning, the vehicle leveling device 3 extends to complete the fixing of the casters 2 and the leveling of the three-wheeled vehicle body 1, laying the foundation for subsequent human-machine collaboration.

[0088] In one implementation, the ball joint 16 is, but is not limited to, connected to the force sensor top plate 7 in a triangular shape. The three branches are, but are not limited to, arranged in a spatially staggered 60° configuration.

[0089] like Figure 4 As shown, three branches are fixed to the upper surface of the vehicle body 1 by branch connecting seats 30. Branch connecting seats 30 are connected to the fourth rotating ball joint 26 and the second rotating ball joint 21. One end of the motor fixing plate 27 is connected to the motor 28, and the other end is fixed to the sleeve 25. The sleeve 25 is equipped with the fourth rotating ball joint 26 and a sliding joint 24 as the output. One end of the pulley 29 is connected to the motor 28, and the other end drives the sliding joint 24 to output. The sliding joint 24 is connected to the sliding joint fixed end 23. The sliding joint fixed end 23 is connected to the RRR lever 20 by the third rotating ball joint 22. The RRR lever 20 is connected to the RS connecting rod 17 by the first rotating ball joint 18. At the same time, the RRR lever 20 is connected to the second rotating ball joint 21.

[0090] RS link 17 is connected to ball joint 16 at one end, and force sensor top plate 7 is installed on ball joint 16.

[0091] In one implementation, the movable pair 24 is driven by a motor 28, which may be connected to the movable pair 24 via a pulley 29 or by direct drive, among other methods.

[0092] like Figure 5 As shown, the mobile parallel robot with high-load pointing and positioning function provided by the present invention includes the following control methods:

[0093] (A) Establish the inverse kinematics of the spatial attitude adjustment device 6, establish the inverse kinematics of the universal wheel 2, and introduce the load compensation coefficient. The calculation formula is:

[0094] ,

[0095] In the formula, The current load weight. The rated load for the space attitude adjustment device 6;

[0096] Calculate the position information of the end-effector 9 at each moment. Solving for the first derivative yields the velocity amplitude, which is then smoothed. Since velocity fluctuates and is subject to disturbances, the velocity information is also smoothed. In the formula The value is 0.05;

[0097] (B) Continue solving for the second derivative to obtain the acceleration amplitude. When there is a disturbance in the acceleration, an acceleration correction factor is introduced for correction.

[0098] ,

[0099] In the formula , For the corrected acceleration, For high load inertial force, The weight of the robot's end effector;

[0100] (C) The human-computer interaction force signal is obtained from the force sensor 8, and the noise is removed by the moving average filtering method to obtain the human-computer interaction force amplitude; the first derivative of the human-computer interaction force amplitude is calculated to obtain the rate of change of the interaction force.

[0101] The denoising formula is:

[0102] ,

[0103] in For the interaction force after noise reduction, Set the sliding window size to 5-10 sampling points. For the sampling period, extract the amplitude of the human-computer interaction force: Perform the first derivative of the interaction force amplitude to obtain the rate of change of the interaction force:

[0104] ,

[0105] Simultaneously, record the fluctuation range of the rate of change of the interaction force:

[0106] This is used for dynamic adjustment of the subsequent reward function;

[0107] (D) State feature extraction and adaptive parameter initialization:

[0108] The amplitude of the filtered velocity information from step (A), the amplitude of the corrected acceleration information from step (B), and the amplitude and rate of change of the human-computer interaction force after denoising from step (C) are extracted as the real-time operating state features of the system. In order to achieve dynamic disturbance rejection under high load, the system adopts an adaptive parameter optimization framework based on reward gradient, initializes the admittance reference parameters of the spatial attitude adjustment device 6 and the omnidirectional wheel 2, namely: the initial values ​​of the damping matrix and the stiffness matrix, and sets the learning rate used to control the step size of parameter updates. , The value is set between 0.9 and 0.95, and is used as the basis for real-time dynamic adjustment of the admittance parameters.

[0109] (E) Hierarchical adaptive admittance control incorporating state characteristics:

[0110] For the spatial attitude adjustment device 6 and the omnidirectional wheel 2, a hierarchical adaptive admittance control structure based on state feedback is designed. First, according to the system's degree of freedom configuration, the end motion extracted in steps (A) and (B) is kinematically decoupled: Z-axis translation, X-axis rotation, and Y-axis rotation components are extracted to construct the motion state vector of the spatial attitude adjustment device. In the formula This is the position / attitude vector of the end effector 9, corresponding to the real-time position / angle of the three degrees of freedom: Z-axis translation, X-axis rotation, and Y-axis rotation. The first derivative corresponds to the velocity / angular velocity of the three degrees of freedom; The second derivative corresponds to the acceleration / angular acceleration of these three degrees of freedom; Corresponding to the interaction force / torque components in these three degrees of freedom;

[0111] Extract the X-axis translation, Y-axis translation, and Z-axis rotation components to construct the motion state vector of the omnidirectional wheel 2 moving base. In the formula This is the position / attitude vector of the omnidirectional wheel, corresponding to the real-time position / angle of the three degrees of freedom: X-axis movement, Y-axis movement, and Z-axis rotation. The first derivative corresponds to the velocity / angular velocity of the three degrees of freedom; The second derivative corresponds to the acceleration / angular acceleration of these three degrees of freedom; These correspond to the interaction force / torque components in these three degrees of freedom.

[0112] Admittance models were then established respectively:

[0113] (a) Establish an admittance model for the Z-axis movement, X-axis rotation, and Y-axis rotation of the spatial attitude adjustment device 6:

[0114] ,in The mass matrix of the spatial attitude adjustment device is 3×3, which is determined by the structural parameters of the mechanism and the high load mass. The corresponding interaction force components extracted in step (C); The time-varying damping matrix and time-varying stiffness matrix are 3×3, respectively. To adapt to high load fluctuation conditions, the adaptive optimization framework constructed in step (D) is used to dynamically update the admittance parameters in real time along the gradient direction of the evaluation function. The update law is:

[0115] ,

[0116] In the formula, The learning step size of the spatial attitude adjustment device 6 was determined through high-load working condition experiments. The evaluation function for the space attitude adjustment device 6 is shown in step (F); during the system initialization phase, the damping matrix and stiffness matrix are preset. The initial values ​​are obtained; within each control cycle of equipment operation, the time-varying damping matrix and time-varying stiffness matrix at the current moment can be obtained. ;

[0117] (b) Establish independent admittance models for the X-axis movement, Y-axis movement, and Z-axis rotation of the omnidirectional wheel 2:

[0118] ,

[0119] In the formula The mass matrix of the omnidirectional wheel is 3×3; These are the time-varying damping matrix and the time-varying stiffness matrix, both 3×3; they adapt to the load disturbance of the omnidirectional wheel 2 and are dynamically updated in real time by an independent adaptive optimization algorithm, with the update law being:

[0120] ,

[0121] In the formula, The learning step size of the omnidirectional wheel 2 is set independently of the learning rate of the spatial attitude adjustment device 6, and is determined through high-load movement experiments. The evaluation function for omnidirectional wheel 2 is shown in step (F); during the system initialization phase, the damping matrix and stiffness matrix are preset. The initial values ​​are obtained; within each control cycle of equipment operation, the time-varying damping matrix and time-varying stiffness matrix at the current moment can be obtained. ;

[0122] (F) Gradient optimization evaluation function design: For the spatial attitude adjustment device 6 and the omnidirectional wheel 2, a system evaluation function for calculating the gradient is designed:

[0123] (a) Evaluation function of spatial attitude adjustment device 6, focusing on the optimization of attitude error and Z-axis displacement error, the specific formula is:

[0124] ,

[0125] In the formula, The weighting coefficients were determined through repeated optimization experiments using high-load pointing positioning. In practical applications, there is a slight emphasis; These are the actual X-axis rotation angle, actual Y-axis rotation angle, and actual Z-axis movement height of the end-effector moving platform 9 of the spatial attitude adjustment device 6, respectively. These are the desired X-axis rotation angle, desired Y-axis rotation angle, and desired Z-axis movement height of the end-effector moving platform 9 of the spatial attitude adjustment device 6, respectively. The sensitivity coefficient is used to adjust the sensitivity of the evaluation function to errors. For the attitude error of the radius, a value of 50 is used. For displacement errors at the millimeter level, a value of 0.03 is used;

[0126] (b) The evaluation function for omnidirectional wheel 2 focuses on optimizing the position error and heading angle error. The specific formula is as follows:

[0127] ,

[0128] In the formula, The weighting coefficients were determined through repeated optimization experiments under high load. In practical applications, there is a slight emphasis; These represent the actual X-axis movement position, actual Y-axis movement position, and actual Z-axis rotation angle of the omnidirectional wheel 2, respectively. These represent the desired X-axis movement position, desired Y-axis movement position, and desired Z-axis rotation angle of the omnidirectional wheel 2, respectively. The sensitivity coefficient is used to adjust the sensitivity of the evaluation function to errors. For the attitude error of the radius, a value of 50 is used. For displacement errors at the millimeter level, a value of 0.03 is used;

[0129] (G) Motor control and drive

[0130] The spatial attitude adjustment device 6 and the omnidirectional wheel 2 adopt a terminal sliding mode control law to ensure that the joint angle tracking error converges within a finite time. The specific formula is as follows:

[0131] ,

[0132] In the formula, To account for the angle tracking error between the spatial attitude adjustment device 6 and the omnidirectional wheel 2, For the desired angle of the spatial attitude adjustment device 6 and the omnidirectional wheel 2, The actual angle acquired by the motor encoder; The coefficients are power coefficients, determined through experimental optimization. To control the gain, it was optimized and determined through high-load operating condition experiments; This is a sign function used to implement disturbance suppression in sliding mode control;

[0133] The formula for converting control torque commands into motor current commands is:

[0134] ,

[0135] In the formula, The torque constant of the motor is... As the reduction ratio of the reducer, the current command is sent to the motor driver to drive the motors of the spatial adjustment device 6 and the universal wheel 2 to rotate.

[0136] As one implementation method, in step (A), the inverse kinematics of the spatial attitude adjustment device 6 is established, specifically as follows:

[0137] Taking the center of the three branches as the origin One branch parallel to the spatial attitude adjustment device 6 is... The axis is perpendicular to the branch and parallel to the upper surface of the vehicle body 1. The axis is perpendicular to the upper surface of the vehicle body 1. Establish a static coordinate system based on the axes. ;

[0138] The origin is taken as the center of the ball joint 16 of the three branches connected to the moving platform 9, namely the ball joint 16 of the first branch 13, the second branch 14, and the third branch 15. Establish a moving coordinate system The axes are all parallel to the static coordinate system. ;

[0139] Let the coordinates of the first rotating spherical joint 18 be... These correspond to the first branch 13, the second branch 14, and the third branch 15, respectively; let the coordinates of the second rotating spherical joint 21 be... These correspond to the first branch 13, the second branch 14, and the third branch 15, respectively; let the coordinates of the ball joint 16 of the three branches relative to the moving platform 9 be . ;

[0140] The offset vector of the moving coordinate system relative to the static coordinate system is ;

[0141] The position vector of the three-branched ball joint 16 in the static coordinate system is: In the formula, The rotation matrix for the moving platform 9.

[0142] The vector from the first rotating ball joint 18 to the ball hinge 16 is The vector from the second rotating spherical joint 21 to the first rotating spherical joint 18 is The closed-loop vector is: ;

[0143] The motion vector of the translating joint 24 is The ratio of the long end to the short end of the RRR lever 20 is... The distance from the second rotating ball joint 21 to the first rotating ball joint 18 is the long end; the distance from the second rotating ball joint 21 to the third rotating ball joint 22 is the short end, with a length of... Then the distance traveled by the traversing joint 24 can be obtained as follows:

[0144] ;

[0145] The distance that the sliding joint 24 needs to move can be converted into the angle that the motor 28 needs to rotate.

[0146] This invention discloses a mobile parallel robot with high-load human-machine collaboration capabilities. Its three-wheeled body has three degrees of freedom: X-axis translation, Y-axis translation, and Z-axis rotation. The spatial attitude adjustment device also has three degrees of freedom: Z-axis translation, X-axis rotation, and Y-axis rotation, and is driven by a lead screw, giving it high rigidity. This invention effectively eliminates mechanical jitter and abrupt changes under high-load conditions by introducing a load compensation coefficient and signal smoothing and denoising. By employing a hierarchical adaptive admittance control and reward gradient optimization framework, combined with a differentiated weight evaluation function, the system's compliance and multi-dimensional collaborative positioning accuracy under complex disturbances are significantly enhanced.

[0147] This invention has the advantages of large load capacity, high positioning accuracy, and convenient human-machine collaboration, and can be applied to assembly scenarios with high-precision pose fine adjustment.

Claims

1. A control method for a mobile parallel robot with high-load pointing and positioning function, characterized in that: Includes the following steps: (A) Establish the inverse mechanism of the spatial attitude adjustment device (6), establish the inverse mechanism of the universal wheel (2), and introduce the load compensation coefficient. Make corrections; Calculate the position information of the end-effector (9) at each moment. Solve the first derivative to obtain the velocity information amplitude, and then perform smoothing processing; (B) Continue to solve the second derivative to obtain the acceleration information amplitude. When there is a disturbance in the acceleration, introduce an acceleration correction factor to correct it. (C) The human-computer interaction force signal is obtained from the force sensor (8), and the noise is removed by the moving average filtering method to obtain the human-computer interaction force amplitude; The first derivative of the human-computer interaction force amplitude is used to obtain the rate of change of the interaction force; (D) State feature extraction and adaptive parameter initialization The above-mentioned velocity information amplitude, acceleration information amplitude, human-computer interaction force amplitude, and interaction force change rate are extracted as the real-time operating status features of the system. Initialize the admittance reference parameters of the space attitude adjustment device (6) and the universal wheel (2), including the initial values ​​of the damping matrix and stiffness matrix, which are used as the basis input for subsequent real-time dynamic adjustment of the admittance parameters; (E) Hierarchical adaptive admittance control combining state characteristics For the spatial attitude adjustment device (6) and the omnidirectional wheel (2), a hierarchical adaptive admittance control structure based on state feedback is designed; according to the system's degree of freedom configuration, the extracted end motion is kinematically decoupled: the Z-axis translation, X-axis rotation, and Y-axis rotation components of the end motion platform are extracted; the X-axis translation, Y-axis translation, and Z-axis rotation components of the end motion platform are extracted; and then admittance models are established respectively. (F) Design of Gradient Optimization Evaluation Function For the spatial attitude adjustment device (6) and the omnidirectional wheel (2), a system evaluation function for calculating the gradient is designed; (G) Motor control and drive The spatial attitude adjustment device (6) and the universal wheel (2) adopt the terminal sliding mode control law to ensure that the joint angle tracking error converges within a finite time. Convert the control torque command into a motor current command; The motor drives the spatial attitude adjustment device (6) and the universal wheel (2) to rotate; The mobile parallel robot with high load pointing and positioning function includes a three-wheeled vehicle body (1), three universal wheels (2) installed on the outer wall of the vehicle body (1), three vehicle body leveling devices (3) installed around the vehicle body (1), a spatial posture adjustment device (6) installed on the upper surface of the vehicle body (1), a force sensor top plate (7) installed on the top of the spatial posture adjustment device (6), a force sensor (8) installed on the upper surface of the force sensor top plate (7), and a moving platform (9) for spatial position adjustment output installed on the top of the force sensor (8).

2. The control method for a mobile parallel robot with high-load pointing and positioning function according to claim 1, characterized in that: In step (A), the inverse kinematics of the spatial attitude adjustment device (6) is established, specifically as follows: Taking the center of the three branches as the origin One branch parallel to the spatial attitude adjustment device (6) is in the following direction. The axis is perpendicular to the branch and parallel to the upper surface of the vehicle body (1). The axis is perpendicular to the upper surface of the vehicle body (1). Establish a static coordinate system based on the axes. ; The origin is taken as the center of the ball joints (16) of the three branches connected to the moving platform (9), namely the ball joints (16) of the first branch (13), the second branch (14), and the third branch (15). Establish a moving coordinate system The axes are all parallel to the static coordinate system. ; Let the coordinates of the first rotating spherical joint (18) be... , respectively corresponding to the first branch (13), the second branch (14), and the third branch (15); let the coordinates of the second rotating spherical joint (21) be . These correspond to the first branch (13), the second branch (14), and the third branch (15), respectively; let the coordinates of the ball joint (16) of the three branches relative to the moving platform (9) be . ; The offset vector of the moving coordinate system relative to the static coordinate system is ; The position vector of the ball joint (16) with three branches in the static coordinate system is: In the formula, The rotation matrix of the moving platform (9); The vector from the first rotating ball joint (18) to the ball hinge (16) is The vector from the second rotating spherical joint (21) to the first rotating spherical joint (18) is The closed-loop vector is: ; The motion vector of the translating joint (24) is The ratio of the long end to the short end of the RRR lever (20) is: The distance from the second rotating ball joint (21) to the first rotating ball joint (18) is the long end; the distance from the second rotating ball joint (21) to the third rotating ball joint (22) is the short end, with a length of Then the moving distance of the traversing joint (24) can be obtained as: ； In the above formula, 0 represents the height of the second rotating ball joint (21) in the Z-axis direction. =0; The distance that the sliding pair (24) needs to move can be converted into the angle that the motor (28) needs to rotate. The spatial posture adjustment device (6) includes three branches with the same structure: a first branch (13), a second branch (14), and a third branch (15). The first branch (13), the second branch (14), and the third branch (15) are arranged in a staggered 60° arrangement between each other. The three branches each include a branch connector (30) fixed on the upper surface of the vehicle body (1), and the upper end of the branch connector (30) is connected to the fourth rotating ball joint (26) and the second rotating ball joint (21); One end of the motor mounting plate (27) is connected to the motor (28), and the other end is fixed to the sleeve (25). The tail of the sleeve (25) is connected to the fourth rotating ball joint (26). The sleeve (25) is equipped with a moving joint (24) as the output. The movable pair (24) is connected to the fixed end (23) of the movable pair. The fixed end (23) of the movable pair is connected to the RRR lever (20) by the third rotating ball joint (22). The RRR lever (20) is connected to the RS link (17) by the first rotating ball joint (18). The RRR lever (20) is connected to the second rotating ball joint (21). The other end of the RS link (17) is connected to the ball joint (16), and the other end of the ball joint (16) is connected to the three vertices of the force sensor top plate (7).

3. The control method for a mobile parallel robot with high-load pointing and positioning function according to claim 1, characterized in that: In step (E), the Z-axis translation, X-axis rotation, and Y-axis rotation components are extracted to construct the motion state vector of the spatial attitude adjustment device. , , , In the formula The position / attitude vector of the end effector (9) corresponds to the real-time position / angle of the three degrees of freedom: Z-axis translation, X-axis rotation, and Y-axis rotation. The first derivative corresponds to the velocity / angular velocity of the three degrees of freedom; The second derivative corresponds to the acceleration / angular acceleration of these three degrees of freedom; Corresponding to the interaction force / torque components in these three degrees of freedom; Extract the X-axis translation, Y-axis translation, and Z-axis rotation components to construct the motion state vector of the omnidirectional wheel (2). , , , In the formula, The position / attitude vector of the omnidirectional wheel (2) corresponds to the real-time position / angle of the three degrees of freedom: X-axis movement, Y-axis movement, and Z-axis rotation. The first derivative corresponds to the velocity / angular velocity of the three degrees of freedom; The second derivative corresponds to the acceleration / angular acceleration of these three degrees of freedom; Corresponding to the interaction force / torque components in these three degrees of freedom; Admittance models were then established respectively: (a) Establish an admittance model for the Z-axis movement, X-axis rotation, and Y-axis rotation of the spatial attitude adjustment device (6): ,in, The mass matrix of the spatial attitude adjustment device (6), i.e., 3×3, is determined by the structural parameters of the mechanism and the high load mass. The corresponding interaction force components were extracted; The time-varying damping matrix and time-varying stiffness matrix are 3×3, respectively. To adapt to high load fluctuation conditions, the adaptive optimization framework constructed in step (D) is used to dynamically update the admittance parameters in real time along the gradient direction of the evaluation function. The update law is: ， In the formula, The learning step size of the spatial attitude adjustment device (6) was determined through high-load working condition experiments. The evaluation function for the spatial attitude adjustment device (6) is shown in step (F); during the system initialization phase, the damping matrix and stiffness matrix are preset. The initial values ​​are obtained; within each control cycle of equipment operation, the time-varying damping matrix and time-varying stiffness matrix at the current moment can be obtained. ; (b) Establish independent admittance models for the X-axis movement, Y-axis movement, and Z-axis rotation of the omnidirectional wheel (2): ， In the formula The mass matrix of the omnidirectional wheel (2); These are the time-varying damping matrix and the time-varying stiffness matrix, respectively; the load disturbance of the omnidirectional wheel (2) is dynamically updated in real time by an independent adaptive optimization algorithm, and the update law is: ， In the formula, The learning step length of the omnidirectional wheel (2) is set independently with the learning rate of the spatial attitude adjustment device (6) and is determined through high-load movement experiments. Let be the evaluation function for the omnidirectional wheel (2); During the system initialization phase, the damping matrix and stiffness matrix are preset. The initial values ​​are obtained; within each control cycle of equipment operation, the time-varying damping matrix and time-varying stiffness matrix at the current moment can be obtained. .

4. The control method for a mobile parallel robot with high-load pointing and positioning function according to claim 1, characterized in that: In step (F), the system evaluation function used to calculate the gradient includes: (a) Spatial attitude adjustment device (6) Evaluation function, focusing on the optimization of attitude error and Z-axis displacement error, the specific formula is: ， In the formula, The weighting coefficients were determined through repeated optimization experiments using high-load pointing positioning. In practical applications, there is a slight emphasis; The actual X-axis rotation angle, actual Y-axis rotation angle, and actual Z-axis movement height of the end moving platform (9) of the spatial attitude adjustment device (6) are respectively. These are the desired X-axis rotation angle, desired Y-axis rotation angle, and desired Z-axis movement height of the end-effector (9) of the spatial attitude adjustment device (6); The sensitivity coefficient is used to adjust the sensitivity of the evaluation function to errors. For the attitude error of the radius, a value of 50 is used. For displacement errors at the millimeter level, a value of 0.03 is used; (b) The evaluation function for the omnidirectional wheel (2) focuses on optimizing the position error and heading angle error. The specific formula is as follows: ， In the formula, The weighting coefficients were determined through repeated optimization experiments under high load. In practical applications, there is a slight emphasis; The actual X-axis movement position, actual Y-axis movement position, and actual Z-axis rotation angle of the omnidirectional wheel (2) are respectively. These are the desired X-axis movement position, desired Y-axis movement position, and desired Z-axis rotation angle of the omnidirectional wheel (2); The sensitivity coefficient is used to adjust the sensitivity of the evaluation function to errors. For the attitude error of the radius, a value of 50 is used. For displacement errors at the millimeter level, a value of 0.03 is used.

5. The control method for a mobile parallel robot with high-load pointing and positioning function according to claim 1, characterized in that: In step (G), the spatial attitude adjustment device (6) and the omnidirectional wheel (2) adopt the terminal sliding mode control law to ensure that the joint angle tracking error converges within a finite time. The specific formula is as follows: ， In the formula, To account for the angle tracking error between the spatial attitude adjustment device (6) and the omnidirectional wheel (2), For the desired angle of the spatial attitude adjustment device (6) and the omnidirectional wheel (2), The actual angle acquired by the motor encoder; The coefficients are power coefficients, determined through experimental optimization. To control the gain, it was optimized and determined through high-load operating condition experiments; This is a sign function used to implement disturbance suppression in sliding mode control; The formula for converting control torque commands into motor current commands is as follows: ， In the formula, The torque constant of the motor is... As the reduction ratio of the reducer, the current command is sent to the motor driver to drive the motor of the spatial posture adjustment device (6) and the universal wheel (2) to rotate.

Images