Strong Generalization Motor Skill Learning and Adaptive Variable Impedance Control for Robots
To address issues such as the complexity of traditional robotic automation in polishing path planning and programming,poor environmental adaptability,and low precision in contact force control,intelligent robotic polishing has become a key approach to over-come these challenges.The core elements of this approach included efficient motion trajectory planning and high-precision contact force control.By using a human-robot physical demonstration of motion trajectories as a reference,research was conducted on robotic polishing skill learning methods.Based on efficient kernel mapping and global stability theory of dynamic systems,a local-global generalization skill model(lgGSM)was developed under the guidance of empirical knowledge.Through human-robot skill transfer,the full degree-of-freedom motion trajectory of the robot's polishing position and posture was efficiently planned.Inspired by the compliant operation mechanism of the human arm,methods for variable impedance control of robotic compliance were studied.The stiffness at the robot's end-effector was esti-mated by electromyography(EMG)signals from the human arm,and a real-time impedance adaptive regulation strategy and contact force compensation mechanism were established to achieve high-precision desired contact force control on complex polishing surfaces.Finally,the robotic polishing system was validated by designing different materials,planes & curves,varying contact forces and polishing paths.The results demonstrate that the method has high trajectory accuracy,good generalization and stability.
万方数据
维普
