Elementary Motion Detection with Contrast Adaptation

Insects display remarkable capabilities for guidance and directed flight based on their visual sense. If duplicated in autonomous flying weapons, these capabilities could greatly enhance their effectiveness. We propose a continuing collaborative effortbetween an insect neurobiologist and VLSI technologist to develop motion processing theory and technology based on insect vision. This effort will apply new findings from insect neurobiology, including work accomplished in Phase I, to the correlationalmodel of the elementary motion detector (EMD), which appears to be the basis for all higher-level visual motion processing in the insect visual system. These new findings show that adaptation of contrast sensitivity plays a crucial role in extending thedynamic range of motion estimation and conferring `velocity constancy,' a response invariance with respect to non-motion-related parameters. New models have been proposed which incorporate adaptation in the EMD, and analog circuit approaches have beendeveloped for the required functions. In this work, we undertake modeling of the adaptive EMD in analog VLSI. We expect the product to be a motion-sensing front end suitable for use in higher-level motion processing, including computation of optical flowand visual detection of small moving targets.