Likelihood Estimation For Slam Using Kinect Device

In the world of robotics, problems of visual SLAM requires an understanding of loop-closure detection and global localization, having said that in order to perform mapping and localization simultaneously we should be able to efficiently recognize an environment that has been previously visited using the current data from our RGBD camera. In this paper we present an online method to recognize and generate information regarding a previously visited place using the visual bag of words model which relies on Bayesian filtering to calculate the probability of loop closure. We would also demonstrate the robustness and effectiveness of our method by real time loop closure detection for an indoor image sequence using Microsoft Kinect camer