Mobile robot localization via classification of multisensor maps

The authors solve the task of mobile robot localization through pattern classification of grid-based maps of important or interesting workspace regions. Each region is represented by registered ultrasound, vision, and infrared sensor grid maps; and feature-level sensor fusion is accomplished by extracting spatial descriptions from these maps. The coarse position of the robot is determined by classifying the map descriptions to recognize the workspace region that a given map represents. Using datasets collected from ten different rooms and ten different doorways in a building, the authors estimate a 94% recognition rate of the rooms and a 98% recognition rate of the doorways. The authors conclude that coarse position estimation in indoor domains is possible through classification of grid-based maps