Performance Analysis and Flight Data Results of Innovative Vision-based Navigation in GPS-denied Environment

B. Polle

Abstract:	Astrium Satellites is developing Vision Based Navigation solutions for safe and accurate autonomous landing on planetary bodies. This transverse technology is being applied to non Space applications as Unmanned Aerial Vehicle (UAV), Miniature Aerial Vehicles (MAVs) and navigation aiding of man piloted helicopters and airplanes. Indeed, all these target applications are highly interested in developing navigation solutions that can work in a GPS-denied environment. A key techniques at the heart of vision-based navigation solutions is the hybridization of IMU and information coming from a mono-camera vision system We will show in this paper how the coupling of a low grade IMU with a low cost 2D-camera can provide a quite accurate dead reckoning navigation estimate at very low cost and mass. As a matter of fact there was an increasing interest in studying optimal fusion techniques of these two types of measurements in the past decade. Astrium Satellites has developed original vision-based navigation solutions with different levels of algorithmic complexity according to the mission requirements and computing resources. The key requirements to these solutions are: (1) The estimated states provided by the navigation solution shall be consistent in order to enable optimal and robust fusion with additional sensors specific to the application, like partial GPS, Doppler Radar, altimeter, etc; (2) The solution shall be still workable when the image complexity and/or size increases. Order O(n) solution are targeted, n being the number of features used in the image (3) The solution shall be robust to the vehicle motion (trajectory, attitude motion) and to the presence of frequent image outliers. This paper presents developed solutions with emphasis on specific theoretical constraints to be solved (observability aspect), as well as on flight test results obtained on aerial vehicles with various types of trajectories and scenes. Comparison with classical SLAM solutions will be addressed. Results will be provided for the critical case of a vertical landing on a plane field highlighting where unobservability issues have to be tackled. The, performances of the developed solutions are presented on three applications. The first study case is the application of vision-based navigation to Lunar Landing, a European Space Agency mission aiming to autonomously, safely, softly and precisely landing a spacecraft at the lunar South Pole by 2018. Then, results on the first flight data acquired by the European Space Agency Planetary Landing Ground Test Facility (PLGTF) will be presented. PLGTF used an unmanned helicopter to fly the NPAL experiment developed under Astrium Satellites lead, consisting of a high performance IMU and a camera based on space environment compatible components, including a Radiation Hardened detector and an image pre-processing FPGA. The trajectory was selected to be as close as possible to a planetary landing trajectory. Finally, flight results of the IRIS experiment dedicated to vision-based navigation will be presented. IRIS experiment developed by Astrium Satellites relies on commercial, off-the shelf sensors, packaged with an acquisition setup having a very limited mass, volume and cost. Several sequences of flight data were acquired with specific scenarios in mind: two “mars-like” quasi-vertical landing trajectories on one side, en-route navigation, aircraft approach on taxiway and unprepared terrain on the other side. It will be shown that the navigation solution allows accurate state estimate even during GPS outage. We demonstrated that Astrium vision-based navigation solutions performed as well as classical SLAM-based solutions, but with significant advantages in term of robustness, consistency and computing resources. This presentation is following a previous presentation (Fischer, Zaunick, Polle, & Kervendal, 2012) made in February 2012 in AAS GN&C Conference in Breckenridge, CO) with more emphasis there on the implementation aspects, the performances for MAVs applications and the latest flight experiment results.
Published in:	Proceedings of the 25th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2012) September 17 - 21, 2012 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	2104 - 2115
Cite this article:	Polle, B., "Performance Analysis and Flight Data Results of Innovative Vision-based Navigation in GPS-denied Environment," Proceedings of the 25th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2012), Nashville, TN, September 2012, pp. 2104-2115.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In