Three Vision-based Behaviors for Self- Positioning a Mobile Robot

This paper presents the analysis of vision-based homing behaviors that provide self-positioning of a mobile robot. The control schemes of the homing behaviors are based on a new approach that describes robot moves in terms of regulation by image features, hence avoiding the usual step of three-dimensional reconstruction of the scene with respect to the camera. Homing behaviors are the key elements of the self-positioning navigation approach we developed for our mobile robot, which basic idea is to represent the robot spatial knowledge in a topological map, where nodes consist in self-positioning sites and edges may be any behavior linking two nodes.