Monte Carlo Localization

In this paper we investigate robot localization with the Augmented Monte Carlo Localization (aMCL) algorithm. The goal of the algorithm is to enable a robot to localize itself in an known world. The map of the environment where the robot has to localize itself must be given to the robot beforehand. In our case the map is constructed by another robot. This map is a so called occupancy grid. An occupancy map contains a value for every location which indicates the probability that this location is occupied by a object such as a wall. We investigate the influence on the performance of the aMCL algorithm when using occupancy maps as input for the aMCL algorithm. We compare the performance of the aMCL algorithm with different levels of uncertainty and two ways of dealing with this uncertainty. We found that the performance of the aMCL algorithm is best when we convert the occupancy map to a binary map by applying a threshold. In that case each location above a certain threshold is considered occupied. This simple method provided the best performance.