Sensing and Power Autonomy for an Insect-Scale Flapping-Wing Robot

Abstract

Advances in mesoscale manufacturing, novel actuation, and control have enabled the design of insect-scale flapping-wing micro air vehicles (MAVs) (mass < 500 mg, wingspan < 5 cm). However, there remain numerous constraints to component technologies that limit their functionality to within controlled flight arenas. Challenges associated with creating autonomous insect-scale flapping wing MAVs are three-fold: 1) many general-purpose sensors and microprocessors do not fit within the size and mass constraints of the vehicles; 2) off-the-shelf actuators do not exist, necessitating an investigation into high-efficiency, low-mass actuation and associated power electronics; and 3) standard power supplies do not have sufficient energy density when reduced to millimeter scales, requiring custom power sources. This thesis describes key contributions towards the design of fully autonomous insect-scale MAVs. Flying insects and larger bird-sized flying robots have a multitude of sensors that provide information about their own state and the surrounding environment. We take inspiration from both to investigate a number of sensors that meet the mass (<50 mg) and power (<10 mW) requirements of these vehicles and can provide proprioceptive feedback during flight. Using the Harvard RoboBee as a platform, we demonstrate the first hovering flights with onboard sensor feedback. Firstly, with a three-axis gyroscope to stabilize the vehicle's attitude during hovering flight, then with a single-axis time-of-flight proximity sensor to stabilize the vehicle's altitude. Finally, we describe the design of a low mass, high efficiency power electronics unit to drive the two piezoelectric bimorph actuators of the RoboBee. Consisting of two bidirectional flyback converters, this custom signal generator is able to amplify a low-voltage input from a power supply to the high-voltage, time-varying unipolar signals necessary to drive the actuators. We then demonstrate the first untethered flight of an insect-scale MAV using this custom signal generator and a photovoltaic array onboard the Harvard RoboBee.