Racing to improve on-line, on-board evolutionary robotics

Abstract

In evolutionary robotics, robot controllers are often evolved in a separate development phase preceding actual deployment - we call this off-line evolution. In on-line evolutionary robotics, by contrast, robot controllers adapt through evolution while the robots perform their proper tasks, not in a separate preliminary phase. In this case, individual robots can contain their own self-sufficient evolutionary algorithm (the encapsulated approach) where individuals are typically evaluated by means of a time sharing scheme: an individual is given the run of the robot for some amount of time and fitness corresponds to the robot's task performance in that period.Racing was originally introduced as a model selection procedure that quickly discards clearly inferior models. We propose and experimentally validate racing as a technique to cut short the evaluation of poor individuals before the regular evaluation period expires. This allows an increase of the number of individuals evaluated per time unit, but it also increases the robot's actual performance by virtue of abandoning controllers that perform inadequately. Our experiments show that racing can improve the performance of robots that adapt their controllers by means of an on-line evolutionary algorithm significantly.