Mobile Robot Localization via Efficient Calibration Technique of a Fixed Remote Camera

Tactical mobility systems have gained focus and interest in several areas including: homeland security, battlefield intelligence, space missions, surveillance systems, etc. Tactical systems work in a very sophisticated environment, and usually they need accuracy and consistency to achieve robust situational awareness. One important aspect of situation awareness is reliable localization of object of interest. This paper describes the design and implementation of an indoor mobile robot localization system using Machine Vision methods, namely, Vision-Based Indoor Positioning System (VIPS). This approach relies on color detection to identify the robot in the environment using a ceiling-mounted camera. Camera calibration model is developed to dynamically estimate the robot position in real time. The camera calibration model is based on environment grid modeling and Euclidian interpolation method to enhance the estimated gross position. Finally, Gaussian zero-phase filter is exploited to refine the localization results. The experiments conducted to evaluate the performance of the proposed system indicate that an accuracy about 6’’ is achievable in complex indoor environment.