Abstract :
We describe an approach to the control of autonomous mobile robots which uses learnt sequences over time of pairs of sensory stimuli and motor actions, so-called sensory-action sequences (SAS), to build an internal model of the world in which the robot is navigating. The hypothesis presented is that the behaviour of an autonomous robot, as with that of an animal, in the learning and performance of a goal-directed, sensory-motor task is determined by the ability of the robot, or animal, to learn and recall spatio-temporal sequences of sensory experiences and associated motor actions. In this paper, we propose a model of the underlying learning and recall process which is biologically-inspired, and is based upon a proposed model of the interaction between certain areas of the animal brain, in particular the prefrontal cortex and certain regions of the limbic system.
