The architecture of a Gaussian mixture Bayes (GMB) robot position estimation system

Abstract

Modelling and reducing uncertainty are two essential problems with mobile robot localisation. In this paper, a new robot position estimator, the Gaussian mixture of Bayes (GMB) which utilises a density estimation technique, is introduced in particular. The proposed system, namely the GMB robot position estimator, which allows a robot's position to be modelled as a probability distribution, and uses Bayes' theorem to reduce the uncertainty of its location. In addition, we describe, in this paper, how our proposed system is capable of dealing with multiple sensors, as well as a single sensor only. Nevertheless, it is known that such multiple sensors could be used to raise more robust than the single sensor, in terms of obtaining accurate estimate over a robot's position. The GMB position estimator mainly consists of four modules such as sonar-based, sensor selection, sensor fusion, and sensor selection improved by combining it with sensor fusion. The proposed system is also illustrated with respect to minimising the uncertainty of a robot's position, using the Nomad200 mobile robot shown in Fig. 1. Eventually, it was found that the proposed system was capable of constraining the position error of the robot by the modularity of the system.