Robot localization and mapping by matching the environmental features from proprioceptive and exteroceptive sensors

Abstract

Robot localization and mapping is an important function of determining the robot pose in an unknown environment. This paper studies the pose estimation by integrating the sensing information both from proprioceptive and exteroceptive sensors. Specifically, the iterative closest point (ICP) algorithm is utilized and modified for characterizing the feature matching of environmental information obtained from encoder (proprioceptive sensor) and laser scanner (exteroceptive sensor). In order to apply the ICP algorithm successfully and efficiently, a searching algorithm is developed to identify the potential common part of two consecutive frames of sensor data. The proposed pose estimation algorithm has been extensively tested in different scenarios with either rich or poor environmental features. Several performance indices are also provided for justify the effectiveness of the algorithm.