Visual SLAM Implementation Using Cloud Robotics

Abstract

Cloud Robotics has gained a lot of momentum with ever growing need for automation. Generally all autonomous robots have the many hardware components to carry out complex computations. This leads to increased weight of robots and increased usage of high voltage batteries or need of frequent recharges of batteries. A minimal hardware architecture on robot can save all of these trouble, which can be achieved by collaborating with cloud. Therefore, cloud robotics aims to shift the computation on cloud, reducing the robotic hardware and promoting sharing of knowledge as well as data amongst robots. Visual SLAM (Simultaneous Localization and Mapping) is the problem of generating map of robotic environment along-with keeping track of the current position of robot within the map. This system proposed here allows the robot to navigate in an environment autonomously based on the predefined set of image templates present in the path of the robot. These images can then be further used to localize the robot in the environment.