A glance at a fly orchestrating a flawless landing on the rim of a teacup, or at a honeybee finding her way back home effortlessly after foraging for nectar several kilometres away, would convince even the most skeptical observer that many insects possess visual systems that are fast, accurate and reliable. Insects are surprisingly adept at coping with their world, despite possessing brains that weigh less than a tenth of a milligram and carry five orders of magnitude fewer neurons than ours do. Studying creatures with "stripped down" nervous systems may be useful in revealing some of the fundamental principles of visual processing and brain function that are common to a number of organisms, including man. On the other hand, it is also conceivable that insects are forced, by the restricted processing power of their brains, to use novel tricks and short-cuts to solve knotty problems of vision and navigation.

Work in our laboratory is revealing that insects display both of these expectations. On the one hand, insects eschew complex stereo-based mechanisms in favour of simpler, image-motion strategies to negotiate narrow gaps, control flight speed, and execute smooth landings, and to obtain a three-dimensional percept of the world. On the other hand, the mechanisms of pattern recognition in the bee seem to be not very different from our own.

The emerging principle is that, for 3-D vision and navigation, insects rely more heavily on motion-based signals than we do. The visual processes by which objects are recognised, classified and distinguished, however, seem to be not very different from our own.

If time permits I shall describe autonomously navigating robots built in our laboratory, based on principles of insect vision and navigation.