Autonomous Agent Map Construction in Unknown Enclosed Environments

MLC-COLT ’94 Robot Learning Workshop, Rutgers, New Brunswick, N.J. ABSTRACT In this research a small, low cost ($250), autonomous mobile robot creates, through exploration and object identification, a landmark map of an unknown room containing obstacles and polygonal furniture-objects possessing walls. The robot, equipped only with inaccurate odometric sensing, short-range IR proximity detection, and long distance acoustic ranging, characterizes or recognizes different polygons objects while contouring them. Experiments indicate the feasibility of the scheme and the sophistication possible even with this minimalist approach.