Vision-Based Navigation II: Error Analysis for a Navigation Algorithm based on Optical-Flow and a Digital Terrain Map

TL;DR

This paper analyzes the errors in a vision-based navigation algorithm that uses optical flow and digital terrain maps, providing a mathematical model to predict pose and motion inaccuracies based on various error sources.

Contribution

It offers a detailed error analysis and closed-form expressions for a navigation algorithm combining optical flow and digital terrain maps, enhancing understanding of accuracy limitations.

Findings

Derived a closed-form expression for navigation error prediction.

Identified key factors affecting navigation accuracy.

Provided a framework for a priori error estimation.

Abstract

The paper deals with the error analysis of a navigation algorithm that uses as input a sequence of images acquired by a moving camera and a Digital Terrain Map (DTM) of the region been imaged by the camera during the motion. The main sources of error are more or less straightforward to identify: camera resolution, structure of the observed terrain and DTM accuracy, field of view and camera trajectory. After characterizing and modeling these error sources in the framework of the CDTM algorithm, a closed form expression for their effect on the pose and motion errors of the camera can be found. The analytic expression provides a priori measurements for the accuracy in terms of the parameters mentioned above.