CAHRA: collision avoidance system for humanoid robot arms with potential field

Abstract

This paper proposes an autonomous collision avoidance system for humanoid robot arms named "CAHRA". There is a difficulty when robot developers make motions of humanoid robot arms, because the right arm and the left arm may collide each other. CAHRA can execute the online collision avoidance, since the computational cost in CAHRA that uses the potential method is very small. An experiment was conducted to confirm the ability of CAHRA which was developed on a humanoid robot "Robovie". The result indicated that CAHRA avoided collision of arms in 97%. Without being nervous about collision avoidance, developers can make robot motions by using CAHRA.