RESEARCH ON MOTOR LEARNING AND CONTROL OF MULTI-DOF BIONIC MANIPULATOR

Motion planning of robotic systems needs to be engineered by professionals, and how to quickly and simply adjust the motion of the manipulator when external tasks change is of major importance to us. We present an action planning method by only using magnetic and inertial measurement unit (MIMU), the entire system consists of a human arm attitude measurement unit and a bionic manipulator control unit. Robotic manipulators can perform fast action learning from the pose data of the operator’s arm instead of complex motion redesigns. The Kalman filter algorithm is used in inertial sensor data fusion, and the fusion of the values recorded from the inertial sensors can be decomposed into the rotation angles of the servos using a rigid body transformation using the Lie group theory. Evaluation tests were performed separately in the LabVIEW platform and on a real robotic system, and the results from the real-time tests show that the method successfully reproduces the movements performed by the operator. © 2024 Acta Press. All rights reserved.