Safe UAV landing: A low-complexity pipeline for surface conditions recognition

As an unmanned aerial vehicle (UAV) navigates autonomously, there are unanticipated occasions, e.g., loss of data provided by the global navigation satellite system, where its mission has to be terminated. In such circumstances, the platform needs to change its navigation mode to landing, so as to protect the system from a possible accident. This process demands the successful selection of the ground surface before the aircraft starts its landing. This paper proposes a low-complexity algorithm for recognizing the suitability of the ground surface based on three laser range-finders, which are mounted on a hybrid vertical take-off and landing (VTOL) fixed-wing UAV. We take advantage of their small size, high precision distance measurements, and operational speed to compute the ground slope and the existence of any obstacles therein. Experiments on a prototype aircraft show that our method can perform robustly and under real-time constraints.