Autonomous ballistic airdrop of objects from a small fixed-wing unmanned aerial vehicle

Autonomous airdrop is a useful basic operation for a fixed-wing unmanned aerial system. Being able to deliver an object to a known target position extends operational range without risking human lives, but is still limited to known delivery locations. If the fixed-wing unmanned aerial vehicle delivering the object could also recognize its target, the system would take one step further in the direction of autonomy. This paper presents a closed-loop autonomous delivery system that uses machine vision to identify a target marked with a distinct colour, calculates the geographical coordinates of the target location and plans a path to a release point, where it delivers the object. Experimental results present a visual target estimator with a mean error distance of 3.4 m and objects delivered with a mean error distance of 5.5 m.