Robotics and Autonomous Systems 37 (2001) 241–260 Learning to select distinctive landmarks for mobile robot navigation Stephen Marsland a , Ulrich Nehmzow b,∗ , Tom Duckett c a Department of Computer Science, University of Manchester, Oxford Road, Manchester M13 9PL, UK b Department of Computer Science, The University of Essex, Colchester CO4 3SQ, UK c Department of Technology, University of ¨ Orebro, Centre for Applied Autonomous Sensor Systems, S-70182 ¨ Orebro, Sweden Received 14 December 2000; received in revised form 1 August 2001 Communicated by F.C.A. Groen Abstract In landmark-based navigation systems for mobile robots, sensory perceptions (e.g., laser or sonar scans) are used to identify the robot’s current location or to construct internal representations, maps, of the robot’s environment. Being based on an external frame of reference (which is not subject to incorrigible drift errors such as those occurring in odometry-based systems), landmark-based robot navigation systems are now widely used in mobile robot applications. The problem that has attracted most attention to date in landmark-based navigation research is the question of how to deal with perceptual aliasing, i.e., perceptual ambiguities. In contrast, what constitutes a good landmark, or how to select landmarks for mapping, is still an open research topic. The usual method of landmark selection is to map perceptions at regular intervals, which has the drawback of being inefficient and possibly missing ‘good’ landmarks that lie between sampling points. In this paper, we present an automatic landmark selection algorithm that allows a mobile robot to select conspicuous landmarks from a continuous stream of sensory perceptions, without any pre-installed knowledge or human intervention during the selection process. This algorithm can be used to make mapping mechanisms more efficient and reliable. Experimental results obtained with two different mobile robots in a range of environments are presented and analysed. © 2001 Published by Elsevier Science B.V. Keywords: Landmark-based navigation; Mobile robots; Kalman filter; Automated landmark selection 1. Introduction 1.1. General considerations — approach For mobile robot navigation over realistic distances, navigation systems based on sensory perceptions or landmarks are usually used. This is because, unlike dead-reckoning methods, perception-based systems ∗ Corresponding author. E-mail addresses: smarsland@cs.man.ac.uk (S. Marsland), udfn@essex.ac.uk (U. Nehmzow), tom.duckett@aass.oru.se (T. Duckett). do not suffer from drift error. However, they are prone to the problem of perceptual aliasing, where a number of perceptions from different parts of an environment look similar. There is another very relevant aspect of landmark- based navigation that has not yet been solved sat- isfactorily. This is the question of what constitutes a ‘good’ landmark, i.e., the question of how land- marks should be selected. So far, this problem has largely been ignored by either logging sensory per- ceptions at regular intervals, or comparing them with pre-installed models of ‘good’ landmarks supplied 0921-8890/01/$ – see front matter © 2001 Published by Elsevier Science B.V. PII:S0921-8890(01)00162-2