Biologically Inspired Feedback Design for Drosophila Flight

Abstract

We use a biologically motivated model of the Drosophila's flight mechanics and sensor processing to design a feedback control scheme to regulate forward flight. The model used for insect flight is the grand unified fly (GUF) [3] simulation consisting of rigid body kinematics, aerodynamic forces and moments, sensory systems, and a 3D environment model. We seek to design a control algorithm that will convert the sensory signals into proper wing beat commands to regulate forward flight. Modulating the wing beat frequency and mean stroke angle produces changes in the flight envelope. The sensory signals consist of estimates of rotational velocity from the haltere organs and translational velocity estimates from visual elementary motion detectors (EMD's) and matched retinal velocity filters. The controller is designed based on a longitudinal model of the flight dynamics. Feedforward commands are generated based on a desired forward velocity. The dynamics are linearized around this operating point and a feedback controller designed to correct deviations from the operating point. The control algorithm is implemented in the GUF simulator and achieves the desired tracking of the forward reference velocities and exhibits biologically realistic responses.