The sensor selection task of the Gaussians mixture Bayes' with regularised EM (GMB-REM) technique in robot position estimation

Modelling and reducing uncertainty are two essential problems of mobile robot localisation. Our previous work has been to develop a robot localisation system, namely the Gaussian mixture Bayes with regularised expectation maximisation (GMB-REM), using a single sensor. It allows a robot position to be modelled as a probability distribution, and uses the Bayes' theorem to reduce the uncertainty of a robot's location. In this paper, a new system, which is enhanced from the GMB-REM system in order to perform a sensor selection task, is introduced. Empirical results show the proposed new system outperforms the GMB-REM system with sonar alone. That is, the new system can deal with multiple sensors and further minimise the average localisation error of the robot by performing the sensor selection task.