Behavioural cloning is a method by which a machine learns control skills through observing what a human controller would do in a certain set of circumstances. More specifically, the machine learns to determine, given the current sensory input, the best of its actions to perform. This paper discusses a robot, situated in the real world, which very quickly learns target pursuit and obstacle avoidance behaviours using this technique. We also discuss the benefits of decision tree learning in the implementation. Key Words: robotics, control, motion, behavioural cloning, imitation