Gesture and End-Effector Trajectory Planning of Robot Manipulator using Human Motion Imitation

• Main Authors: Yu-Cheng Liu, 劉育成
• Other Authors: 林顯易
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/c7nury

碩士 === 國立臺北科技大學 === 自動化科技研究所 === 100 === In order to operate a redundant robot arm to accomplish the path planning, and intuitively control its posture simultaneously in on-line programming, incorporating the human motion is a useful and flexible option. We have introduced the human motion imitation, Inverse Kinematics problem and mapping algorithm in our thesis. First, we discuss motion capture and data processing. We introduce how to translate human motion to a robot arm, and discuss the advantages and drawbacks of the proposed methods. Secondly, we discuss two appropriate methods to address the inverse kinematics problem, namely the Inverse Jacobian with Multi-tasking method, and Forward and Backward Reaching Inverse Kinematics (FABRIK) method. We propose how to augment the Inverse Jacobin with Multi-tasking method to solve two of the disadvantages. The main idea to solve those disadvantages is that exchanging the task priority while the algorithm is in progress results in a conflict between the primary task and the lower priority task. Thirdly, we have designed a mapping algorithm to control a robot to follow the trajectory of human motion on-line. Finally, in the simulations and experiments, we have compared those algorithms we have discussed: accuracy, similarity and computation time. The results show that a robot or model can imitate the human motion institutively and flexibly on-line.