Augmenting State Estimates with Multiple Camera Visual Measurements
Olivier Toupet, James D. Paduano, Robert Panish, Raymond J. Sedwick, Emilio Frazzoli · AIAA Infotech@Aerospace 2007 Conference and Exhibit · 2007
This paper provides recent results on the use of multiple cameras in a coupled vision/ inertial navigation framework for aerospace vehicle state estimation. The framework allows data from any number of cameras to be systematically combined. Multi-camera configurations including binocular and orthogonal cameras are considered. Kalman filtering concepts are applied for fusing vision inputs and inertial navigation information (gyros and accelerometers). The resulting architectures are useful for on-board state estimation in GPS-denied environments. Methods have been tested in a unique development framework that includes direct feature motion simulation as well as feature motion processed from high-resolution three-dimensional simulations.