Frontal Target Approaching and Height Control of Micro Air Vehicles using Multi-perspective Optical Flow Observables

This paper proposes a novel strategy for forward distance and height estimation for Micro Air Vehicle (MAV) navigation using optical flow observables from multiple perspectives. This strategy, inspired by flying insects like honeybees using wide catadioptric imaging systems, leverages the observation that flow patterns diverge near the focus of expansion and appear nearly uniform when the view is directed away from it during target approaching. An extended Kalman filter that leverages these two key observables—flow divergence and ventral flow—is introduced to estimate both the forward distance and height of MAVs during target approaching. To ensure safe and efficient approaching, a constant flow divergence control method is integrated, which results in an exponential decay in both forward distance and velocity to zero upon reaching the target. The proposed strategy was implemented and validated through numerical simulations and real-world flight tests with an MAV, which demonstrated its effectiveness and efficiency in the estimation. Experiments tracking various flow divergence setpoints showcased the method’s robustness in accurately estimating forward distance and height even under varying approaching dynamics and noisy measurements. These estimates could enable online path planning, real-time decision-making, and accurate height control.