Object traversing by monocular UAV in outdoor environment

In search and monitoring tasks, object traversing scenarios are more challenging than avoidance and tracking. In this paper, we propose an object perception and path planning algorithm which can detect hollow objects and calculate a traversing path in real-time. Meanwhile, a general vision unmanned aerial vehicle (UAV) system framework is designed to implement the proposed algorithm. A series of fundamental but effective algorithms are utilized in combination to detect hollow objects, and then the triangulation is applied to calculate the 3D position of the object. In order to improve the detection accuracy, a path planning and motion control algorithm is designed to enhance the safety and stability during traversing. The actual experimental results show that the UAV system is robust and feasible.