Positioning by tree detection sensor and dead reckoning for outdoor navigation of a mobile robot

Abstract

We propose a positioning method for outdoor navigation of a mobile robot, by fusing dead reckoning and the tree detection sensor which consists of sonar and vision. A street lined with trees is assumed to be the mobile robot's outdoor work space. In this environment, trees are good landmarks for robot's position estimation. This paper describes a method for robot position estimation by fusion of dead reckoning and tree detection sensor based on maximum likelihood estimation at first. Then, the method for the detection of tree using the sensor system with sonar and vision mounted in one body is described. At last, the experimental results of self-guidance with the experimental autonomous mobile robot "YAMABICO" is presented. The result shows the effectiveness of the our method for outdoor navigation of the mobile robot.<>