Robust local localization for indoor environments with uneven floors and inaccurate maps

A robust local localization algorithm is presented for use in indoor environments that processes line segments extracted from a 2D laser range finder. To better handle real-world environments the proposed solution is designed to be tolerant of uneven floors, common map errors, and objects not found in the map, while using minimal computational resources. The primary contributions of this work are: 1) novel motion estimation equations which are numerically stable under all geometric conditions, 2) methods for handling uneven floors, and 3) explicit design and testing for stability under common map errors.