We study what navigational capabilities a robot may exhibit by exploiting primitive (but robust) visual information related to the bearing angles of landmarks in its visual field. Preliminary work on this subject started with the study of the robot homing problem but has recently been extended to more general robot navigation behaviors. You may find interesting to visit the related page of Kostas Bekris and Lydia Kavraki at Rice University, USA.
Download video with experiments with angle-based navigation in various interesting cases (collinear landmarks, dynamic environment, robot kidnapping, natural landmarks, etc). The perceptual support of angle-based navigation is provided by our home-built multiple targets tracking algorithm.
The electronic versions of the above publications can be downloaded from my publications page.