Integrated indoor navigation system for aerial vehicle using visual odometry and artificial landmark matching

摘要

Robustness and efficiency of indoor navigation at high rates is crucial for aerial vehicle to perform maneuvering action and fast exploration. In recent years, optical flow sensors have shown its effectiveness on indoor navigation of mobile robot. However the fatal problem of optical flow's drifting can only be solved by building a map using SLAM (Simultaneously Localization and Mapping), which would drop the rate to nearly 30Hz. Navigation using artificial landmarks is the earliest visual navigation method implement on mobile robot, which, also has a limitation when the number of visible artificial landmark is insufficient. This paper explore the feasibility of combination of navigation from optical flow and artificial landmarks which will complement mutually. The integrated navigation system is experimented on the aerial vehicle exploring the laboratory, comparing with the optical flow method and landmarks method separately.