An Approach to Labelled Indoor Mapping using SLAM and Object Detection

In this paper, an in-house developed robot, named as ExploDroid, is introduced with details of its construction, mapping, object-detection implementation and functional demonstration. It is developed using the Robot Operating System (ROS) platform. The proposed application of this robot is to obtain a labelled 2D map of an unknown static indoor environment using Simultaneous Localization and Mapping (SLAM) algorithm, for creation of a map and also for object-detection in it. The objects would then be appropriately labelled. The end result is a 2D map of the environment, with labelled objects. The SLAM algorithm is implemented using the open-source Gmapping package, which is integrated into the ROS Framework. Besides, it is also used alongside the RViz Visualization tool for viewing the maps created. By employing low-cost hardware, the cost of building such a platform has been made quite economical while retaining the precision of the map obtained. Experimental trials in real world environments yielded accurate and verifiable object-detection and labelling results.