Migratory birds-inspired navigation system for unmanned aerial vehicles

摘要

Migration birds are able to navigate themselves during a long-distance journey without getting lost. They actually achieve just what is being sought for in the field of Unmanned Aerial Vehicles (UAVs): long-term autonomous navigation. This paper proposes an approach that combines the migration birds' sense principles with Micro-Electro-Mechanical System (MEMS) sensors to estimate UAVs position within GPS-denied environments. Camera, orientation and web-based maps (such as Google/Baidu Maps) are chosen to simulate the birds' localization cues: vision, earth magnetic field and mental maps. The visual odometry, Particle Filter theories are used in the proposed approach to integrate multiple sensor measurements. Real flying experiments are conducted both in indoor and outdoor environments. The results validate that the proposed migration-inspired visual odometry system can estimate the UAV localization effectively.