Webcam configurations for ground texture visual servo

Increase in accessibility of low cost webcams has lead the way for hobbyists and researchers to include affordable, yet information rich, visual sensors for areas such as machine vision. This work investigates the practical considerations of using a floor pointing visual sensor for visual servo applications. Several conditions and characteristics of both the webcam and the ground texture tracking are explored to provide efficient use of the camera. Using the discovered characteristics, several image restoration filter algorithms are found to improve the signal to noise ratio with good efficiency and efficacy in ground texture tracking applications. The findings here form the basis for a localization algorithm for a mobile robot platform.