University of Maryland researchers working with Aurora Flight Sciences on a bio-inspired, vision-based navigation system for micro air vehicles have sent us this video of one of their laboratory experiments. It shows a micro-helo navigating its way along a 'corridor' in the lab using the visual phenomenon known as "optic flow" to avoid colliding with the walls.
Video: UMD Autonomous Vehicle Laboratory
Optic flow is generated by the vehicle's motion relative to its surroundings. In insects, specialized neurons in the visual system, called tangential cells, are tuned to recognize patterns in this wide-field optic flow and signal the nervous system. UMD's Prof Sean Humbert says his lab has mathematically replicated this process and found it's a quick and easy way to extract relative proximity and speed information.
In the demo above, a sensor on the micro-helo is measuring optic flow in azimuth and extracting relative proximity, both in heading and laterally, by using actual pattern sensitivities recorded in insects. Aurora and UMD believe this bio-inspired technique could allow MAVs to navigate down city streets. Because the wide-field optic flow sensor does not have the resolution to detect small obstacles, it will be augmented with sonar, which would allow the MAV to detect and dodge poles and wires using bat-inspired echolocation.