Tree-select Trial and Error Algorithm for Adaptation to Failures of Redundant Manipulators

Abstract

In this paper, we introduce a novel algorithm we call "Tree-select Trial and Error" (TTE) that can help a redundant robotic arm to adapt to failures that might occur during its functioning. This algorithm proposes a new search strategy allowing the robot to generate new behaviours, rather than being restricted to the previously learned ones, in order to compensate damage effects. Thus, learning suitable behaviours become an online process that promotes the robot's knowledge. The simulation and the experimental results on a planar manipulator with six actuators proved the effectiveness of our algorithm in helping the robot to reach a large area of its workspace in a few number of trials despite the different damage scenarios.