Autonomous mobile robot localization using Markov decision algorithm

Abstract

Localization is one of the fundamental problems of autonomous robots. This problem is related to determining the position of a mobile robot from sensor data. However, the mostly solution to localization problems is passive, meaning they do not manage the opportunity to control the robot's effectors during localization. This paper proposes an active localization approach, which is Markov localization. Markov localization provides rational criteria for setting the robot's motion direction (exploration) and determining the pointing direction of the sensors to localize the robot most efficiently. Furthermore, it can deal with noisy sensors and approximative world models, which is far more beneficial. After the demonstration in a structured environment of 5×5 boxes area, Markov can find the actuator position within four steps. The error probability reading shows a result of 3/100 for the highest and 1/100 for the lowest. The corresponding error measurement is discussed. © 2023 Author(s).