Progress Towards Untethered Autonomous Flight of Northeastern University Aerobat

TL;DR

This paper reports progress in developing autonomous untethered flight for Northeastern University's Aerobat, a lightweight bio-inspired flapping wing micro aerial vehicle, focusing on state estimation and control challenges with limited onboard computation.

Contribution

It presents novel approaches to perception, state estimation, and control for a lightweight, resource-constrained flapping wing MAV aiming for untethered autonomous flight.

Findings

Achieved initial autonomous flight capabilities.

Developed perception methods suitable for limited onboard resources.

Identified key challenges in control and estimation for small bio-inspired MAVs.

Abstract

State estimation and control is a well-studied problem in conventional aerial vehicles such as multi-rotors. But multi-rotors, while versatile, are not suitable for all applications. Due to turbulent airflow from ground effects, multi-rotors cannot fly in confined spaces. Flapping wing micro aerial vehicles have gained research interest in recent years due to their lightweight structure and ability to fly in tight spaces. Further, their soft deformable wings also make them relatively safer to fly around humans. This thesis will describe the progress made towards developing state estimation and controls on Northeastern University's Aerobat, a bio-inspired flapping wing micro aerial vehicle, with the goal of achieving untethered autonomous flight. Aerobat has a total weight of about 40g and an additional payload capacity of 40g, precluding the use of large processors or heavy sensors. With limited computation resources, this report discusses the challenges in achieving perception on such a platform and the steps taken towards untethered autonomous flight.