Evolutionary Robotics aims to apply evolutionary computation techniques to evolve the overall design or controllers, or both, for real and simulated autonomous robots. Numerous challenges exist that relate for example to (1) how to distribute intelligence between the body and brain (2) how to evolve robust behaviours that are tolerant to changes in the environment or the robot itself either in an individual robot or in a swarm of robots (3) the role of diversity in morphology/behaviour in contributing to robust behavior (4) how to ''learn to learn'', i.e. evolve learning strategies that provide rapid learning in an adaptive environment. A PhD candidate could focus on one or more of these areas: although some of the work would be undertaken in simulation, the student will also have access to a physical swarm of 50 heterogeneous robots.