Vision-Aided Navigation Using a Tightly-Coupled INS and Predictive Rendering Techniques

J. Beich, M. Hanson, B. Pugsley, J.D. Slack, and M. Veth

Abstract:	Relative navigation using a tightly-coupled Inertial Navigation System (INS) and vision system in a three-dimensional, GPS-denied environment is a topic of great interest for both defense-related and civilian research. Past vision-based efforts in weaponry have primarily treated the navigation problem as flat and two-dimensional and generally offered no way to improve performance when it is most desired - at endgame. A unique algorithm is developed that compares images taken by an on-board camera system with predictively-rendered views of a three-dimensional target environment model to determine navigation error. The navigation error is then combined with INS mechanization via a modified nonlinear Unscented Kalman Filter (UKF) to compute optimal navigation state estimates. The algorithm is dubbed Vision-Aided Navigation using Statistical Predictive Rendering (VANSPR). Flight test was conducted to collect images, position truth data, and raw inertial data to practically test the algorithm in a “real world” environment. Truth data from the flight test was used to conduct simulations to finish development of the Kalman filter mechanization. Results indicate that the Vision-Aided Using Statistical Predictive Rendering (VANSPR) algorithm has the potential to be a viable method of aiding an inertial-only navigation system in the context of tactical weaponry. A 63% improvement over a free-inertial system was demonstrated using a preliminary ”onesize-fits-all” automated measurement model. Sensitivity analysis of significant parameters show much better results can be obtained by altering measurement model values for specific runs. It is anticipated that a more robust measurement automated model will be developed in this ongoing research that will yield improved consistent results across diverse data samples.
Published in:	Proceedings of the 24th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2011) September 20 - 23, 2011 Oregon Convention Center, Portland, Oregon Portland, OR
Pages:	3557 - 3570
Cite this article:	Beich, J., Hanson, M., Pugsley, B., Slack, J.D., Veth, M., "Vision-Aided Navigation Using a Tightly-Coupled INS and Predictive Rendering Techniques," Proceedings of the 24th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2011), Portland, OR, September 2011, pp. 3557-3570.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In