We present an approach to solving the problem of autonomous map building and path planning for a mobile agent. Unlike conventional approaches, we feel that this meets the requirements of a general purpose system which includes, robustness, autonomy, flexibility and scalability. Underpinning this technique is new representation of shape using visual features, based on geometric co-occurance which we call the Corner-centred Geometric Histogram (CGH). Results are presented which demonstrate that these histograms can be used for both localisation and navigation purposes. This is due to their rich description of a scene and predictable response to changes in viewing position.
Keywords:
Autonomous Mobile Robot, Map Building, Geometric Histogram, View-Based
