Practical feature-based simultaneous localization and mapping using sonar data

A practical feature-based SLAM is proposed for applications in indoor environment using sonar data. Line features are extracted from groups of sonar data using hybrid Hough transform and least square (HTLS) algorithm. To meet the real-time requirement, an efficient implementation of HTLS algorithm based on slide window is introduced. Besides, the standard formulation of EKF-SLAM is extended to 2D-line case where the practical plant and observation model is constructed and then employed. To illustrate the performance of the proposed algorithm, experiment results on a Pioneer mobile robot are provided, which demonstrate that the constructed SLAM technique exhibits superior practicality and simplicity.