How to Evolve Autonomous Robots: Different Approaches in Evolutionary Robotics

摘要

number of parts or modules within the system; rather, it scales with the number of possible interactions between parts and modules.A methodology for evolving the control systems of autonomous robots has not yet been well established.In this paper we will show different examples of applications of evolutionary robotics to real robots by describing three different approaches to develop neural controllers for mobile robots.In all the experiments described real robots are involved and are indeed the ultimate means of evaluating the success and the results of the procedures employed.Each approach will be compared with the others and the relative advantages and drawbacks will be discussed.Last, but not least, we will try to tackle a few important issues related to the design of the hardware and of the evolutionary conditions in which the control system of the autonomous agent should evolve.(b) autonomous robots interact with an external environment and, therefore, the way in which they behave in the environment determines the stimuli they will receive in input (Parisi, Cecconi, and Nolfi, 1990).Each motor action has two different effects: (1) it determines how well the system performs with respect to the given task; (2) it determines the next input stimuli which will be perceived by the system (this last point strongly affects the success or the failure of a sequence of actions).Determining the correct motor action that the system should perform in order to experience good input stimuli, is thus extremely difficult because any motor action may have long term consequences.Also, the choice of a given motor action is often the result of the previous sequence of actions.A final source of uncertainty in the design of the system is the fact that often the interaction between the system and the environment is not perfectly known in advance.