The fuzzy logic of visuomotor control

Abstract

Biological sensorimotor control is characterized by the use of signals from large numbers of sensors, monitoring numerous variables. Among these are the exteroceptive signals from the eyes and ears. Many of the sensory signals are under efferent control, and the motor responses they evoke, whether at a simple reflex level or routed through the higher centres, appear to be task and context dependent. In technology the analysis and management of multiple-input, multiple-output systems clearly exceed the capabilities of classical servo control theory. In this commentary, new types of control system based on conditional logic are discussed in relation to the-rules animals use to control movement. It is argued that the concepts of fuzzy logic control provide a useful and 'biologically compatible' way of describing sensorimotor behaviour. An example is given of a robotic device under fuzzy control, in which behaviours are selected according to a visual assessment of motor task and context. Each behaviour is associated with a small subset of rules relating specific sensory variables to specific motor-actions. The rule-based approach is also discussed in relation to neurophysiological theories regarding the interneuronal control of locomotion, including the recently adduced 'parliamentary principle'. The analysis and classification of behaviours and rules is seen as a useful preliminary to the future study of interneuronal systems.