Many generic position-estimation algorithms are vulnerable to ambiguity introduced by nonunique landmarks. Also, the available high-dimensional image data is not fully used when these techniques are extended to vision-based localization. This paper presents the landmark matching, triangulation, reconstruction, and comparison (LTRQ global localization algorithm, which is reasonably immune to ambiguous landmark matches. It extracts natural landmarks for the (rough) matching stage before generating the list of possible position estimates through triangulation. Reconstruction and comparison then rank the possible estimates. The LTRC algorithm has been implemented using an interpreted language, onto a robot equipped with a panoramic vision system. Empirical data shows remarkable improvement in accuracy when compared with the established random sample consensus method. LTRC is also robust against inaccurate map data.

Simultaneous localisation and map-building (SLAM) can be considered as a combined state and parameter estimation problem. Instead of using extended Kalman filtering, a more flexible Sequential Monte Carlo method is considered. Multiple generic particle filters are initialised to estimate the robot and obstacle positions concurrently. Simulation results based on a simple robot environment, which represents obstacles by line segments, indicate the feasibility of the proposed method.

A panoramic image based localisation system has been implemented on a B21r mobile robot and tested in a heavily cluttered environment. It takes vertical object edges as natural landmarks and extracts a one-dimensional token sequence from the input image before matching with reference image sequences. Matched points are then triangulated to give a set of position estimates, and the best result is taken as the current robot position. Selection criteria and methods for the reference image sites are also discussed, including the use of generalised Voronoi vertices.