Search

CN-122008200-A - Control method and data acquisition method for teleoperation of robot

CN122008200ACN 122008200 ACN122008200 ACN 122008200ACN-122008200-A

Abstract

The application provides a control method and a data acquisition method for teleoperation of a robot, wherein the control method comprises the steps that an operator wears an exoskeleton to acquire an arm joint angle and a hand joint angle in real time; forward kinematics calculation is carried out based on the arm joint angle to obtain the pose of the tail end of the wrist, and meanwhile, the finger pose is obtained through hand pose self-adaptive calculation; and mapping and inverse kinematics resolving the pose of the tail end of the wrist, controlling and driving the mechanical arm to move, and mapping the pose of the finger to the mechanical arm to execute control. The application realizes high-precision and stability control by exoskeleton collecting angle information, forward and backward kinematic calculation, mapping and the like.

Inventors

  • ZHU HAOHUA
  • ZHOU CHEN

Assignees

  • 浙江灵巧智能科技有限公司

Dates

Publication Date
20260512
Application Date
20260127

Claims (10)

  1. 1. A control method for teleoperation of a robot, comprising: the operator wears the exoskeleton to acquire the arm joint angle and the hand joint angle in real time; forward kinematics calculation is carried out based on the arm joint angles to obtain the pose of the tail end of the wrist, and meanwhile, hand pose self-adaptive calculation is carried out on the hand joint angles to obtain the pose of the finger; And mapping and inverse kinematics resolving the pose of the tail end of the wrist, controlling and driving the mechanical arm to move, and mapping the pose of the finger to the mechanical arm to execute control.
  2. 2. The control method for teleoperation of a robot according to claim 1, wherein the forward kinematic solution based on the joint angle, to obtain a wrist end pose, comprises: taking a base coordinate system as a reference origin, and traversing each connecting rod joint step by step according to the kinematic topological structure of the exoskeleton; for each joint, calculating the local pose transformation of the joint relative to the father node according to the type and the input real-time joint angle; multiplying the pose of the father node by the relative local pose transformation of the current joint through homogeneous transformation recursion on the plum cluster SE (3) to obtain the accumulated pose of the child node under the global coordinate system; and repeating the above process until the tail end of the motion chain formed by the connecting rods outputs the pose of the tail end of the wrist.
  3. 3. The method for controlling a teleoperation of a robot according to claim 1, wherein the adaptively resolving the hand gesture of the hand joint angle to obtain a finger gesture comprises: In an initial calibration stage, based on sensor data acquired in a wearing state of a wearer, identifying a reference gesture of a finger, and constructing an individualized hand kinematics model by combining a space position of a key point and a priori proportion of human kinematics; In a dynamic operation stage, the real-time sensing information and the individualized hand kinematics model are utilized to jointly solve the joint angles of all the knuckles through a geometric constraint and kinematic inversion method, and the finger gesture is reconstructed.
  4. 4. A control method for robotic teleoperation according to claim 1, characterized in that said mapping of the wrist end pose comprises: mapping the pose of the wrist end into a safe working space of a mechanical arm of the machine side by using a motion space mapping coefficient to obtain a target position of the wrist end of the machine side, wherein the target position is specifically as follows: X r =C r +r.(X h C h ) Wherein X r is the calculated target position of the wrist end of the mechanical arm on the machine side, X h is the position of the wrist end on the human side, C h is the initial position when the human hand starts teleoperation, C r is the initial position when the mechanical arm starts teleoperation, r is the adjustable motion space mapping coefficient, and the range of values is (0, 1).
  5. 5. The method according to claim 4, wherein the performing inverse kinematics calculation on the pose of the wrist end and controlling the driving mechanical arm to move in combination with PD comprises: Calculating the current position and posture of the tail end of the wrist of the mechanical arm by adopting forward kinematics, and comparing the current position and posture with the target position of the tail end of the wrist to obtain a space error; Based on the space error, constructing an optimization problem of a joint speed stage by combining a jacobian matrix and a damping least square strategy of the mechanical arm; aiming at the optimization problem, solving and gradually adjusting the joint angle in an iterative mode to approach the joint configuration meeting the target pose; evaluating convergence in each iteration, if the pose error reaches the preset precision, accepting the current solution, otherwise rejecting the solution result, maintaining the last iteration state, and continuing to solve; and finally, outputting the converged joint angle instruction and the execution state thereof to a bottom controller of the mechanical arm, and taking the joint angle instruction and the execution state as target input of PD control to realize stable track of the track.
  6. 6. The method for controlling teleoperation of a robot of claim 1, further comprising feeding back haptic information to the wearer after the telerobotic manipulator has completed the tele-action, comprising: acquiring a pressure sensor output signal integrated by a mechanical finger tip, and detecting a normal contact force received by the pressure sensor output signal; Dividing the force value range for applying force to the grabbing target into five grades, and mapping the force value range to the corresponding feedback grade according to the detected normal contact force; And converting the feedback level into the vibration intensity of the corresponding finger vibration motor on the exoskeleton glove, and transmitting the tactile sensation to an operator in a continuous vibration mode.
  7. 7. The control method for teleoperation of a robot according to claim 6, further comprising triggering a vibration alarm to alert an operator to adjust the magnitude of the applied force when an overload or abnormal load is detected, in particular: monitoring any one of the following parameters in real time: (a) The calculated hand final output vibration feedback amount based on the tactile sensor reading and the current sensor reading, wherein the vibration feedback amount is determined by: ; Wherein sensors [ i ] comprises a touch sensor reading and a current sensor reading, gamma is a mapping coefficient of current to vibration intensity, and n is the number of hand touch sensors and current sensors; (b) Deviation q of current actual position q current of mechanical finger from target position q goal error =∣q goal q current ∣; If any of the parameters exceeds a preset threshold, judging that the stress blocking or the load abnormality occurs, and triggering a high-intensity vibration alarm, wherein the alarm adopts a pulse type or full-frequency vibration mode to be different from the conventional tactile feedback.
  8. 8. A data acquisition method for teleoperation of a robot, comprising: at the human end, the control method of any one of claims 1-7 is adopted to perform teleoperation control on the robot at the machine end, and the mechanical arm are driven to execute corresponding movements; during control and execution of the movement, one or more of the following data are synchronously acquired and recorded: a control instruction which is generated by the control method and is derived from a human end and a corresponding timestamp thereof; The sensing data from the machine end sensor comprises joint states of the mechanical arm and environmental perception information; based on the data with the aligned time stamps, a multi-mode teleoperation data set with time sequence consistency is constructed.
  9. 9. A terminal comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor is operable to perform the method of any of claims 1-8 when the program is executed by the processor.
  10. 10. A computer readable storage medium having stored thereon a computer program, which when executed by a processor is operable to perform the method of any of claims 1-8.

Description

Control method and data acquisition method for teleoperation of robot Technical Field The application relates to the field of teleoperation of robots, in particular to a control method and a data acquisition method for teleoperation of robots. Background In recent years, with the rapid development of smart operation technology of robots, various intelligent control models based on deep learning and reinforcement learning are developed successively. The training of these models is highly dependent on a large amount of high-quality demonstration data, including both virtual data in the simulation environment and real machine operation data covering the real robot performing tasks, thus placing an urgent need for large-scale, high-precision robot operation data. Although some disclosed robot training data sets exist at present and provide rich action samples, the covered mechanical arm configuration and end effector types are limited, the requirements of diversified and complicated smart operation tasks are difficult to meet, and particularly, the cross-platform migration and generalization capabilities are obviously limited. To obtain more practical and representative robot operation data, researchers have widely adopted teleoperation (Teleoperation) methods to collect human demonstration actions. Through enabling an operator to remotely control the robot to finish various fine tasks, high-fidelity behavior track data can be generated for subsequent imitation learning and strategy training. At present, teleoperation systems mainly depend on various data acquisition devices to realize, and common schemes comprise: the vision-based motion capture system has lower cost, but is easily influenced by factors such as hand shielding, illumination change and the like; The IMU glove can sense the bending gesture of the finger more accurately, but has insufficient precision in global positioning and space gesture estimation; VR handle equipment, which has strong portability but is limited by dependence of hand tracking frequency and environment and poor stability; although the exoskeleton type device is slightly complicated to wear, stable and accurate measurement of hand movement can be realized by virtue of the rigid connecting rod structure and the high-precision angle sensor. However, in actual remote operation data acquisition, the accuracy and stability of control directly determine the quality of the acquired demonstration data, thereby affecting the learning effect and task success rate of the robot. Especially in subtle tasks requiring millimeter-scale operation precision, such as grasping, assembling, rotating, etc., minor motion distortions may cause learning failures or execution deviations. Disclosure of Invention In view of the shortcomings/drawbacks of the prior art, it is an object of the present application to provide a control method and a data acquisition method for teleoperation of a robot. In a first aspect of the present application, there is provided a control method for teleoperation of a robot, comprising: the operator wears the exoskeleton to acquire the arm joint angle and the hand joint angle in real time; forward kinematics calculation is carried out based on the arm joint angles to obtain the pose of the tail end of the wrist, and meanwhile, hand pose self-adaptive calculation is carried out on the hand joint angles to obtain the pose of the finger; And mapping and inverse kinematics resolving the pose of the tail end of the wrist, controlling and driving the mechanical arm to move, and mapping the pose of the finger to the mechanical arm to execute control. Optionally, the performing forward kinematic solution based on the joint angle to obtain a wrist end pose includes: taking a base coordinate system as a reference origin, and traversing each connecting rod joint step by step according to the kinematic topological structure of the exoskeleton; for each joint, calculating the local pose transformation of the joint relative to the father node according to the type and the input real-time joint angle; multiplying the pose of the father node by the relative local pose transformation of the current joint through homogeneous transformation recursion on the plum cluster SE (3) to obtain the accumulated pose of the child node under the global coordinate system; and repeating the above process until the tail end of the motion chain formed by the connecting rods outputs the pose of the tail end of the wrist. Optionally, the performing hand gesture adaptive calculation on the hand joint angle to obtain a finger gesture includes: In an initial calibration stage, based on sensor data acquired in a wearing state of a wearer, identifying a reference gesture of a finger, and constructing an individualized hand kinematics model by combining a space position of a key point and a priori proportion of human kinematics; In a dynamic operation stage, the real-time sensing information and the individualized ha