Multi sensor data fusion for 6D pose estimation and 3D underground mine mapping using autonomous mobile robot. (3rd April 2017)

• Record Type: Journal Article
• Authors: Ghosh, Dibyendu; Samanta, Biswajit; Chakravarty, Debashish
• Is Part Of: International journal of image and data fusion. Volume 8:Number 2(2017)
• Page Start: 173
• Page End: 187
• Publication Date: 2017-04-03
• Subjects: Multi-sensor data fusion -- EKF -- 6D pose estimation -- scan matching -- SLAM -- underground mine mapping; Image processing -- Periodicals; Multisensor data fusion -- Periodicals; Multisensor data fusion; Periodicals; 621.36705
• DOI: 10.1080/19479832.2016.1226966
• Languages: English
• ISSNs: 1947-9832

Abstract

ABSTRACT: Simultaneous localisation and mapping by using a robot is indeed a challenging and critically important problem for an accurate 3D underground mine mapping. Using 6D pose information of a robot ( x, y and z coordinates and the yaw, pitch and roll angles), the robot vehicle can be tracked, which is very crucial for an accurate mine mapping. The estimation of 6D robot pose on the ground surface is easy and generally accomplished by either the Global Positioning System (GPS) or by putting some artificial visual landmarks. However, in the underground mining environment, the estimation of pose becomes difficult due to the unavailability of the GPS signal and the identification of landmarks (as surface wall of the underground mine is rough). In this paper, an attempt has been made to construct a 3D underground mine map using a prototype mine robot. For this purpose, a strategy for the underground mine robot pose estimation is devised with the help of multi-sensor data fusion techniques. A shaft-encoder (wheel odometry), Inertial Measurement Unit (IMU) and a rotated 2D Laser Scanner is used to improve and rectify the 6D robot pose as well as 3D mine map. An autonomous prototype robot is built up and allowed to navigate in a real underground mine gallery to validate our algorithm. Generated 3D map has been compared with the actual map available. The outputs of this study are highlighted in this paper.