Line-based incremental map building using infrared sensor ring

Geometric representation of the environment, known as mapping, plays an important role in mobile robotics as it support various tasks such as localization, path planning and motion control. Line models represent a popular approach to represent the geometric features of the environment. The mapping problem gives rise to a variety of solutions for using different exteroceptive sensors (sonar, infrared, laser, vision, etc.). One of the possible choices is to use an infrared sensor (IR sensor). These devices are inexpensive, practical and widely available. Infrared sensors are largely used in obstacle avoiding processes because they are fast and cheap, and require only simple signal processing. Although the collision avoidance routine has received good attention for many years, the infrared feature mapping has been neglected. A problem with this kind of sensor is its short range (about one meter). This problem forces to an incremental mapping approach, so the robot has to move and concurrently model the environment. In this paper we present an IR sensor able to measure distances based on the light that is back-scattered from objects. This IR sensor is used for line-based map building purposes, so we also present a methodology for line extraction, range data clustering and line segmentation. The experimental test has been carried out in a long corridor, and results show the usefulness of the IR sensor and the methodology we propose for feature extraction and line-based mapping processes.