Year: 2024
Type: conference-paper
Abstract: Ability to navigate in the unknown GPS-denied environments, such as harsh or indoor urban scenarios, without the need of aids like satellite navigation remains a challenge for unmanned aircraft, while insects have been doing this effortlessly with a tiny brain for millions of years. Insects posses neurons inside their brain that are sensitive to visual motion known as optic flow, and it is well known that insects rely on this visual cue for their navigation tasks. We present a novel bio-inspired active sensing technique that has been observed in bees for flight control. Our technique uses controlled and calibrated lateral motion to generate translational optic flow in the ventral field of view to estimate the altitude of the aircraft. Optic flow vectors are estimated from images captured by a downward-facing onboard camera mounted on a gimbal. A complete closed-loop system was simulated in Microsoft AirSim with realistic scenery generated in the Unreal Engine. The obtained simulation results show that the aircraft can achieve closed-loop control from taking off, reaching the desired altitude and maintaining it with lateral sinusoidal motion.
Institutions: UNSW Sydney, UNSW Canberra, The University of Queensland
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4408199485