An improved indoor localization system for mobile robots based on landmarks on the ceiling

Abstract

Accurate and reliable localization is a basic requirement for a mobile robot to execute tasks including docking and transportation. A common approach to solve the problem is to use artificial landmarks. In this paper, an indoor localization system using passive landmarks and an infrared sensor is introduced. However, there are several types of estimation error in this localization system. To solve these problems, a sensor model is built after the analysis to the properties of the infrared sensor. We use the sensor model to incrementally place landmark on the robot's path. It aims to reduce the number of landmarks with the variance guarantee to hold. Then we introduce how to create a global coordinate system. A data fusion method based on the sensor model is also used when landmark is on the view edge of the sensor to improve location accuracy. Experiments performed demonstrate the feasibility of improved localization system.