An indoor navigation algorithm incorporating representation of Quasi-Static Environmental Object and glass surface detection using LRF sensor

Abstract

To autonomous robot systems, performing tasks, such as navigation a mobile robot in indoor environments should be able to cope with state changes of many environment objects, such as a door. This will involve many key technologies such as mapping, real time state observation and action decision. In this paper, we define the concept of Quasi-Static Environment Object (QSO) which represents state-changeable object from mapping level, and implement it in current robotic map system with QSO data structure. A navigation algorithm with QSO state representation and states based navigation decision is developed for allowing a robot can flexibly modify its route while observed door in its original shortest route is closed. Additionally, a method with general type LRF sensors for glass QSO object detection is proposed by using reflection characteristics of glass and pre-registered QSO map data. This lets our proposed extended navigation algorithm can be applied to most of indoor environments including many recent buildings with transparent glass doors and glass walls which are invisible in conventional SLAM based mapping methods. Several experiments are provided for illustrating the validity of the proposed methods.