Closed-loop performance of a proportional controller for visual stabilization using a fly-robot interface

Abstract

The blowfly Calliphora is the model of choice for studying sensori-motor control principles common in biological systems. We present a fly-robot interface where the neural activity of an identified visual interneuron is used to control the angular velocity of a rotating robot. By placing the robot on a rotating turn-table in a visual arena, we use the fly-robot interface to quantify the dynamics and performance of a proportional controller in a closed-loop visual stabilization system. The properties of the system were characterized for both step and frequency responses. We analysed the data using a performance index based on the input-output energy dissipated by the controller. Our results suggest that the optimal strategy for the fly to minimize the visual slip speed would be to tune the closed-loop gain to the angular velocity and angular acceleration of the input stimuli. The design principles discovered by reverse-engineering sensori-motor control in to develop the next generation of autonomous robots and smart sensors.