Algorithm for Autonomous Landing
- Tuesday, 01 March 2011
Because of their small size, high maneuverability, and easy deployment, micro aerial vehicles (MAVs) are used for a wide variety of both civilian and military missions. One of their current drawbacks is the vast array of sensors
(such as GPS, altimeter, radar, and the like) required to make a landing. Due to the MAV’s small payload size, this is a major concern.
The key to this approach is to calculate time-to-collision based on some image on the ground. By holding the angular velocity constant, horizontal speed decreases linearly with the height, resulting in a smooth landing. Mathematical proofs show that even with actuator saturation or modeling/measurement uncertainties, MAVs can land safely. Landings of this nature may have a higher velocity than is desirable, but this can be compensated for by a cushioning or dampening system, or by using a system of legs to grab onto a surface.
Such a monocular camera system can increase vehicle payload size (or correspondingly reduce vehicle size), increase speed of descent, and guarantee a safe landing by directly correlating speed to height from the ground.