A Striding Algorithm to Prevent Humanoid Robots from Falling due to External Forces

• Main Authors: Cho-Li Pao, 鮑卓立
• Other Authors: Wei-Chung Teng
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/49957276407784295767

碩士 === 國立臺灣科技大學 === 資訊工程系 === 103 === To interact with human, humanoid robots should be able to maintain its balance against external forces. Thus, this research studied preferred striding patterns to respond to external forces. These patterns should keep the robots balanced without altering their front directions. In this work, the robot's acceleration and pressure sensing data are used to determine the magnitude and direction of external force. These knowledge is then mapped to the robot's step size and angle by applying LIPM (Linear Inverted Pendulum Module) and law of conservation energy. Since the angle of every rotational joint has its limit, 3 patterns are developed responding to forces of different magnitude. For verification, Webots simulator is utilized to simulate the cases when external forces from 8 different directions work upon NAO robot. The experimental results show that the developed patterns successfully make the humanoid robot to regain its balance state.