Gesture reconfiguration from joint failure using genetic algorithm

Abstract

Humanoid robot consists of many degrees of freedom to perform various tasks in the same way as human. However, most robots lost its abilities when the joint failure problem occurs. In order to keep its basic function when the robot suffered joint failures, we proposed the algorithm which applied basic concepts of genetic algorithm optimization for generating new configuration from the remaining joints so that the robot can continue to perform its task. In this paper, we showed the experimental results which applied the proposed gesture reconfiguration using genetic algorithm optimization method to the semi-humanoid reception robot named Namo when the joint failure occurred. This robot has two 7 dofs arms which are used for performing emblematic gestures when interacts with human. The joint failure causes the robot not to be able to perform these predefined emblematic gestures. The proposed algorithm will generate a set of new gestures from the remaining working joints, which can still be recognized by human as the original gestures.