Adaptive tracking for a mobile robot

A novel technique for adaptive tracking in indoor lighting environments based on Chebyshev's theorem is presented. The technique is used to recover the region corresponding to an artificial landmark accurately and efficiently through a sequence of images. Accurate region segmentation is the first step in determining the position of a mobile robot relative to a landmark. Nonadaptive region tracking techniques are susceptible to even small variations in indoor illumination and as a result may return degraded regions. An adaptive feedforward technique is necessary to combat this degradation, which is measured in terms of preservation of the centroid, region size, and visual erosion. This technique has been tested successfully using black and white images acquired from a mobile robot. Demonstrations of the adaptive tracking technique working in conjunction with the move-to-goal and follow-the-leader behaviors on the mobile robot are presented.<<ETX>>