Abstract
In recent years, monocular vision has been considered as an effective Unmanned Aerial Vehicle (UAV) onboard obstacle perception solution with applications to Sense and Avoid (SAA). However, with the limitations of monocular optical measurement, monocular vision based obstacle localization ability is insufficient for collision avoidance. Therefore, an obstacle collision avoidance trajectory planning scheme is proposed in this paper with considerations of monocular optical measurement characteristic. Firstly, two obstacle localization modes are defined in this paper to assist with obstacle collision avoidance, namely relative range based Mode 1 and relative angle based Mode 2. Obstacle localization observability of Mode 2 is analyzed, and coordinate systems for obstacle localization are constructed based on the analysis results. Given coordinate systems, Orthogonal Iteration (OI) is further adopted for obstacle localization. Secondly, due to the lack of global knowledge caused by monocular vision based localization capability, a rolling horizon based safe trajectory planning method is presented. In each time segment, the trajectory is optimized with considerations of: 1) objective functions including minimum trajectory length, elapsed time and energy consumption; 2) constraints including Mode 1 and 2 based obstacle collision avoidance constrains and UAV flight constraints. Finally, simulation results indicate the proposed obstacle collision avoidance trajectory planning scheme enhances UAV safety level and can achieve favorable performance when compared with geometric based obstacle collision avoidance and collision avoidance with global knowledge.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call