Many of the tasks posed to robots in the course of research are rather arbitrary, e.g. catch a ball, walk over uneven ground, don't get stuck. However most are justified by a background assumption that they will be needed for carrying out culturally salient tasks such as cleaning, exploring Mars or acting as a personal servant. A related or alternative justification is that these are kinds of things animals (including humans) can do.
A large proportion of work in robotics draws on standard engineering techniques and control theory. This tends to conceptualise behaviour as something the robot must produce despite the effects of the external environment. Some skepticism about whether this is how biological systems (sensorimotor task performers par excellence) actually work has led an alternative approach. This emphasises that behaviour is the outcome of interactions with the environment and seeks to exploit that interaction in designing controllers. However, though this has led to some very interesting machines, it is still debatable whether it is more effective than the traditional approach, especially in terms of real applications.
I will discuss various examples to illustrate the current state of robot development, and where it might be going. I will then look more specifically at issues in relating the study of robots to the study of animals (including humans).
Dr. Barbara Webb
AI Research Group
Psychology Department
University of Nottingham