A H∞ controller synthesis framework for vision based obstacle avoidance applications

This paper presents an insect inspired visual navigation technique that combines spatial decomposition of instantaneous optic flow patterns with a H_∞ loop shaping approach to extract local proximity and velocity information in a cluttered, obstacle laden environment. Loop shaping provides a controller synthesis framework that can be used to account for environment structural uncertainties and gusts. Closed loop stability and safe navigation behavior with this approach are demonstrated in simulation on an autonomous helicopter miro-air-vehicle (MAV) for an urban-like environment subjected to gusts, that has the added advantage of computational simplicity which satisfies the stringent size, bandwidth and power constraints imposed by MAVs.