A sensor fusion-based optimization method for indoor localization

Abstract

Indoor differential-driven mobile robots play a crucial role in industrial applications such as intelligent inspection and smart transportation. In scenarios with uneven road friction and/or sharp turn, challenges remain for accurate robot localization in indoor environment. When using encoders of the wheel odometry as the sole basis, the robot will lead to significant odometry drift and inaccurate localization. To optimize the wheel odometry data, this paper presents an internal and external combined sensor fusion method for laser simultaneous localization and mapping (SLAM) system. Experimental results demonstrate that the proposed method significantly reduces cumulative localization error by comparing with method using solely wheel odometry. It also effectively corrects an average translational odometry error caused by inherent wheel odometry drift.