Evolving Robot Morphology and Control

Most robotic approaches begin with a fixed robot hardware design and then experiment with control structures. We take a different approach that considers both the robot hardware and control structure as variables in the evolutionary process. This paper reports the results of experiments which explore the placement of sensors and effectors around the perimeter of a simulated agent's body, and the Neural Network (NNets) that controls them. 1 Introduction Evolutionary algorithms have been used in the design of robot controllers for some time and with great success. Almost all work on evolving autonomous agents and robots has focused on evolution of the control structure, often a NNet. This work has taken as a basic assumption that the agent body is immutable. This is a consequence of the flexibility of software over that of hardware: in the past it has not been feasible to explore the space of robot morphology. In contrast to this, the course of natural evolution shows a history of body,..