A Review on Method of Obstacle Detection and Avoidance Systems for Unmanned Aerial Vehicles: Monocular Camera

Abstract

Unmanned Aerial Vehicles (UAVs) or known as drone is widely use by human as high risk and time-consuming task can be tackle by UAV. The usage of UAV includes the movement and path of the UAV to execute given tasks. However, any obstacles will restrict the movement of UAV thus various method is discovered and undergoes research specifically to detect and avoid any upcoming obstacles. This review paper will review vision sensor that include monocular and stereo camera sensor. Numerous researchers focus more on using monocular camera sensor as the main image input for image processing as it give benefits in terms of payload and power consume to the operational hours of the UAV. Therefore, this paper will focus on reviewing obstacle detection and avoidance method that use mainly monocular camera with respect to image cue to avoid obstacles. The method to extract cue from the image for this paper review will fall in the boundary of appearance-based, motion-based, depth-based, expansion-based of the input image. As a results from all techniques, texture-less, narrow, and small moving objects can be assumed as hard to detect and challenging that need to be tackled in future research particularly important for aerial vehicles.