On localising an unknown mobile robot

Abstract

This paper presents an exteroceptive based localisation system that makes the unconventional assumption that a robot's locomotion mechanism, and its physical interaction with the environment, is unknown. As a consequence, the system functions independently of dead-reckoning estimates and can therefore be considered a self-contained sensing unit with the ability to position an arbitrary robotic platform. The proposed system uses exteroceptive based perception to form hypotheses about the robot's absolute pose over time. The hypotheses are derived by algorithmically solving the kidnapped robot problem, within a simultaneous localisation and mapping (SLAM) framework. Preliminary results were gathered in real-time using a scanning laser rangefinder, mounted on a skid steering mobile robot, in an outdoor environment. This novel approach to localisation was found to be robust and versatile; with possibly a commercial bent, supported by its generality.