Real time visual SLAM using cloud computing

Simultaneous localization and map-building (SLAM) continues to draw considerable attention in the robotics community due to the advantages it can offer in building autonomous robots. However, few approaches to this problem scale up to handle a large number of landmarks present in real environments. The processing resource requirement to carry out SLAM in real time can be quite high. In this paper we present a novel system which employs resources provided by infrastructure as a service (IaaS) and parallelism for effective processing. Using private cloud infrastructure employing virtualized resources based on MPI the task of global map building is performed in real time simultaneously carrying out loop detection and bundle adjustment for indoor environments. Through implementation in various challenging environments with moving obstacles, visually homogeneous areas having few features, regions with large changes in lighting and relatively fast camera motion we demonstrate our system to be one which is effective as well as robust.