Bearing-Only SLAM with Indistinguishable Landmarks

Abstract

To navigate successfully, a mobile robot must be able to estimate the spatial relationships of the objects of interest in its environment. The main advantage
of a bearing-only SLAM system is that it requires only a cheap vision sensor for navigation.
In this paper, we present a bearing-only SLAM method for a 2D environment with visually
indistinguishable landmarks (Unknown Data Association). When the frame rate of the vision sensor is high, it is easy to track the landmark correspondences. When the frame rate is low, we estimate the landmark positions by computing the most likely hypothesis based on bearing measurements.
A bearing-only SLAM system that employs this method does not require any other sensors such as range sensors or wheel encoders. Due to its low cost, this system is ideally suited for domestic robots such as autonomous lawnmowers and vacuum cleaners.