Indoor 3D Object Model Obtained using Data Fusion from Laser Sensor and Digital Camera on a Mobile Robot

This paper proposes a framework to build a 3D model for an indoor object from partial data using a mobile robot. The purpose of the paper is to obtain photorealistic 3D models for objects from raw mobile laser scanning data and digital camera information (to improve the appearance information). The proposed method integrates and analyses the relationships between the visual data and raw mobile laser scanning data. This method is capable to recover the complete density range and the shape of the investigated object. In order to obtain 3D mapping using the laser sensor system (called Lidar), this is positioned to a vertical scan orientation and the tilt actuator provides a rotation movement to the Lidar in order to perform a 3D scan. In this paper, the mobile system acquires 3D Lidar Data only while it is stationary. A 3D map of an object is created by acquiring overlapping scans from multiple locations and viewpoints using local co-ordinates for each scan and merging the scan into a single global co-ordinate system. Video registration is used to texture the scan points in order to provide photo realistic 3D maps. Some techniques are also implemented to improve the resolution of the Lidar Scan with video data. Experiments on real-world data are given to illustrate the suitability of this approach.