Bayesian Learning Experiments with a Khepera Robot

This paper presents a new robotic programming environment b ased on the probability calculus. We show how reactive behaviours, like obs tacle avoidance, contour following, or even light following, can be programmed and learn d by the Khepera with our system. We further demonstrate that behaviours can be combi ned either by programmation or learning. A homing behaviour is thus obtained by combinin g obstacle avoidance and light following.