Simultaneous localization and sampled environment mapping

Abstract

Simultaneous localization and map building is a key issue to ensure the mobile robot move in an unknown environment autonomously. A hot topic of SLAM is how to build a map describing the complex environment. This paper presents a new SLAM algorithm using the sampled environment map, which describes the environment in detail, rather than represent the environment with a small number of geometric parameters. The proposed method segments measurements into primitive objects and fits them with implicit polynomials. Algebraic distances or orthogonal distances are then considered as new measurements, which are used to update the whole state. A method considering geometric constraints is presented to remove redundant environment samples from the SEM. The algorithm's main merits are its compactness and adaptability. Simulation and experimental results demonstrate the efficiency of our algorithm.