Odometry-less localization in tunnel-like environments

Localization of mobile robots in straight and smooth environments, such as tunnels, mines or pipes, is an extremely difficult task due to the lack of recognizable features. Most of the laser-based localization algorithms, in fact, rely on these features to compute the position by matching successive laser readings. To solve this problem, we propose a technique for localizing a robot in such environments using exclusively a LIDAR sensor and a peculiar movement pattern. This allows, by means of trivial trigonometric calculations, computing the longitudinal position with good precision, avoiding the necessity of feature-matching. The algorithm has been implemented and tested through simulations and real-world experiments. The results are compared with odometry and the GMapping algorithm, proving that the proposed scheme yields interesting results in terms of accuracy in this kind of environments.