Optimal placement of range-only beacons for mobile robot localisation

Abstract

The ability of an agent to self-localise is crucial to any autonomous task where mobility is required. A common set of techniques solving the localisation problem involve the deployment of active beacons or landmarks, which eliminate problems related to landmark detection and association. The use of beacons providing range-only estimates using time-of-flight measurements is one such approach. Here, range measurements are used in trilateration or range-only SLAM algorithms to provide an accurate measure of a robot’s position. Unfortunately, the potential error in a position estimate is related to the relative geometry of the beacons, and poorly placed beacons can result in extremely inaccurate location estimates. This paper presents an optimisation technique for finding optimal beacon positions, so as to minimise the mean positional uncertainty in a given environment. Our work shows that this approach represents an improvement on previous approaches because the resultant uncertainty map can be used as a heuristic to improve path planning algorithms.