Complex behaviour generation on autonomous robots: a case study

This paper proposes a framework for developing distributed architectures aimed at complex behavior generation in autonomous robots. Without engaging radical "mentalistic" or behaviorist ideas, the system developed is a useful tool to explore scalability, modularity and complexity issues in building autonomous robots. To show its validity, a case study is presented in which a mobile robot endowed with a vision system performs the "go to target" task. Several topics, including the identification of classes of problems that might constitute building blocks of more general forms of intelligence in autonomous robots, the use of space-variant vision or the problem of "getting stuck" for a robot, are addressed.