A Performance Metric for Mobile Robot Localization

Abstract

This paper focus on the problem of how to measure in a reproducible way the localization precision of a mobile robot. In particular localization algorithms that match the classic prediction-correction model are considered. We propose a performance metric based on the formalization of the error sources that affect the pose estimation error. Performance results of a localization algorithm for a real mobile robot are presented. This metric fulfils at the same time the following properties: 1) to effectively measure the estimation error of a pose estimation algorithm, 2) to be reproducible, 3) to clearly separate the contribution of the correction part from the prediction part of the algorithm, and 4) to make easy the algorithm performance analysis respect to the great number of influencing factors. The proposed metric allows the validation and evaluation of a localization algorithm in a systematic and standard way, reducing workload and design time.