|Abstract:||Vision measurements are inherently angular measurements, leaving room for improvement in some situations by augmenting vision based navigation algorithms with attitude inputs. Commercial off-the-shelf autopilots are increasingly popular for use in UAVs and often generate low precision attitude solution. This paper describes UAV flight tests which evaluated the performance of two different kinds of vision navigation algorithms, each of which was implemented with and without autopilot aiding. The first flight test looked at absolute positioning using automatic feature matching with a feature database and a PnP solution. The second flight test looked at two-frame visual odometry. Both had a Pixhawk autopilot available to aid the vision navigation algorithm, and performance was evaluated with and without such aiding. Results show that the attitude aiding was a significant benefit in both the absolute positioning and visual odometry cases. The reasons for this are explained, and lessons for future flight testing of similar systems are presented.|
Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017)
September 25 - 29, 2017
Oregon Convention Center
|Pages:||2321 - 2331|
|Cite this article:||
Leishman, Robert C., Gray, Jeremy, Raquet, John, "Utilization of UAV Autopilots in Vision-Based Alternative Navigation," Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017), Portland, Oregon, September 2017, pp. 2321-2331.
ION Members/Non-Members: 1 Download Credit