The flight performance of insects depends critically on rapid and robust sensory-motor control. My laboratory is attempting to determine the means by which the fly’s nervous system integrates visual, olfactory, and mechanosensory input to appropriately modify motor output. By tracking freely-flying flies within a large arena it is possible to reconstruct what a fly ‘sees’ as it moves through a structured visual landscape. The flight paths of fruit flies consist of straight sequences interspersed with rapid saccadic turns, during which the animal changes its heading. During the straight sequences the animals ignore certain features of optic flow, flying instead under the control of modified wings called halteres that function as gyroscopes. While the temporal and spatial distribution of saccades may appear quite stochastic, after reconstructing the visual input it is possible to link nearly every saccade turn to a point when the rate of expansion within one visual field passed a critical threshold. After each saccade, flies attempt to stabilize sharp contrast edges resulting in the straight flight sequences. As they approach an object, however, the asymmetrically looming edges elicits first a decrease in forward velocity and then a rapid collision-avoidance saccade. Chemical odors appear to influence flight behavior, not by overriding optomotor control, but rather by altering the gains and thresholds of visual control circuits. This clever fusion of olfactory and visual information produces a robust and efficient search algorithm and should serve as a useful model for control systems in autonomous vehicles.
This seminar is being presented as part of CDS 101, Principles of Feedback and Control. If you would like to be added to the mailing list to receive future CDS 101 seminar announcements, you can subscribe to the cds101-announce mailing list.