Self-adjusting locomotion on a partially broken-down quadrupedal biomorphic robot by evolutionary algorithms

Biomorphic robots have become an interesting topic recently. These robots can achieve certain goals that wheel-based robots cannot. The biomorphic robots usually have more joints and more legs. However, the more motors on a robot the more risk that one of them might break down at an unexpected moment. Self-adjusting locomotion ability can be a help to make a partially dysfunctional biomorphic robot move. Online evolutionary algorithm is a promising way to achieve such a task. The robot receives feedback from the environment as to the fitness for its evolutionary goal. In this paper, we adopt the PSO (particle swarm optimization) algorithm as our online evolutionary algorithm and test it on a partially broken-down quadrupedal biomorphic robot. The experimental results show that the robot can adjust its actions to move even when one leg is removed.