Computer Vision Final Project
Robot Localization Using an Omnicamera
Fidicuals vs PCA

Amin Atrash
Michael Kaess
Ernest Merrill


Purpose

The purpose of this project is to use vision techniques for localization of a robot in an indoor enviornment using an omnicam. One technique will involve a finding fiducials in the enviornment and using a Kalman filter to localize. A second technique will involve using a Principle Component Analysis(PCA) of the images to localize. These are both based on previous work by [Dellaert].

Equipment
Finding Fiducials
Predicting Fiducials and the Kalman Filter
PCA
References
Code


Last modified: Thu Dec 6 09:50:31 EST 2001