Share Email Print
cover

Proceedings Paper

Optic flow aided navigation and 3D scene reconstruction
Author(s): Malcolm Rollason
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

An important enabler for low cost airborne systems is the ability to exploit low cost inertial instruments. An Inertial Navigation System (INS) can provide a navigation solution, when GPS is denied, by integrating measurements from inertial sensors. However, the gyrometer and accelerometer biases of low cost inertial sensors cause compound errors in the integrated navigation solution. This paper describes experiments to establish whether (and to what extent) the navigation solution can be aided by fusing measurements from an on-board video camera with measurements from the inertial sensors. The primary aim of the work was to establish whether optic flow aided navigation is beneficial even when the 3D structure within the observed scene is unknown. A further aim was to investigate whether an INS can help to infer 3D scene content from video. Experiments with both real and synthetic data have been conducted. Real data was collected using an AR Parrot quadrotor. Empirical results illustrate that optic flow provides a useful aid to navigation even when the 3D structure of the observed scene is not known. With optic flow aiding of the INS, the computed trajectory is consistent with the true camera motion, whereas the unaided INS yields a rapidly increasing position error (the data represents ~40 seconds, after which the unaided INS is ~50 metres in error and has passed through the ground). The results of the Monte Carlo simulation concur with the empirical result. Position errors, which grow as a quadratic function of time when unaided, are substantially checked by the availability of optic flow measurements.

Paper Details

Date Published: 15 October 2013
PDF: 15 pages
Proc. SPIE 8897, Electro-Optical Remote Sensing, Photonic Technologies, and Applications VII; and Military Applications in Hyperspectral Imaging and High Spatial Resolution Sensing, 88970H (15 October 2013); doi: 10.1117/12.2027498
Show Author Affiliations
Malcolm Rollason, QinetiQ Ltd. (United Kingdom)


Published in SPIE Proceedings Vol. 8897:
Electro-Optical Remote Sensing, Photonic Technologies, and Applications VII; and Military Applications in Hyperspectral Imaging and High Spatial Resolution Sensing
Gary W. Kamerman; Gary J. Bishop; John D. Gonglewski; Ove K. Steinvall, Editor(s)

© SPIE. Terms of Use
Back to Top