Initial position estimation of a mobile robot with a laser range finder by differential evolution

Abstract

For autonomous mobile robots, it is essential to have a capability of estimating its own position and orientation based on some sensing data. This problem is called ‘localisation’ in robotics societies. This paper treats the portion of localisation problems at the beginning of the movement. In this paper, we consider the case when the local sensor attached to the mobile robot is only a laser range finder. The unknown variables to be estimated are the 2-D position and the orientation. This problem can be solved by general algorithms for non-linear optimisation, and among them, we apply differential evolution. As the fitness function, we use the mixture probability density function that can meet various conditions of scanning targets. The usefulness of the proposed method is demonstrated through experiments.