Colloquium "Implicit Observation Model for Vision Aided Inertial Navigation of Aerial Vehicles Using Single Camera Vector Observations"

Abstract:
In many aerial vehicles operational navigation systems the usage of inertial sensors alone experiences slow divergence in position, velocity and attitude parameters. This divergence is often compensated by employing some additional source of navigation information external to Inertial Navigation (IN). Vision measurements, in particular single camera measurements, are the propitious choice because of their low cost, low size and power requirements and compactness. We address the IN divergence and propose novel solution that provides a source of reliable aiding information to IN. Using the single camera unit direction vector measurements and with no a priori knowledge of the environment we show that, while moving in vicinity of a map point, it is possible to constrain the IN position, velocity and attitude divergence. Using the discovered fact that the relative range rate is a function of the velocity, attitude and the camera measurements, novel implicit observation model is developed. Since the relative range rate is quantity not measurable in the navigation process we propose an approach to augment it in the vehicle state and to form augmented state vector. When the vehicle is moving in vicinity of an observed map point the camera measurements are used to constrain the divergence of the position, velocity and attitude and to update the relative range rate.

CV: 
Vasko Sazdovski received the M.Sc. degree in electrical engineering from the Institute of Automation and Systems Engineering, Faculty of Electrical Engineering and Information Technologies, University of Ss. Cyril and Methodius, Skopje, R. Macedonia in 2010 (field of mobile robotics). From 2006 until 2012 studied for a part time PhD at the Centre for Autonomous Systems, Department for Informatics and Systems Engineering, Cranfield Defence and Security, Cranfield University, Swindon, United Kingdom (field of Vision ided Inertial Navigation). His research interests are in Robotics, automation and systems engineering. He is particularly interested in: sensor and data fusion, stochastic filtering and estimation techniques, stochastic control techniques.