Vision Based Surface Slope Estimation for Unmanned Aerial Vehicle Perching

Unmanned aerial vehicles are applied to many different fields such as surveillance, search, and monitoring. However, a critical issue for unmanned aerial vehicles is that they suffer from short flight time. To address this, perching becomes a new necessary capability for unmanned aerial vehicles. However, before perching on the desired surface, usually the orientation of the UAV needs to be adjusted to make the perching mechanism to firmly attach to the surface. In this paper, a vision algorithm is introduced to estimate the surface slope of the perching object. Equipped with a distance sensor and a monocular camera, the surface slopes in both X and Y directions can be estimated simultaneously. A bunch of experiments with different slope combinations are carried out. Combined with a Kalman Filter, the experiment results show this algorithm is much better compared with the previous algorithms especially when the main movement of the unmanned aerial vehicle is along the camera optical axis.