Görsel eşzamanlı haritalama ve konumlandırma probleminin performansını aykırı gözlemleri optik akı yardımıyla eleyerek artırma

Simultaneous Localization and Mapping (SLAM) for mobile robots has been oneof the challenging problems for the robotics community. Extensive study of thisproblem in recent years has somewhat saturated the theoretical and practicalbackground on this topic. Within last few years, researches on SLAM have beenheaded towards Visual SLAM, in which camera is used as the primary sensor.Superior to many SLAM application run with planar robots, VSLAM allows us toestimate the 3D model of the environment and 6-DOF pose of the robot.Being applied to robotics only recently, VSLAM still has a lot of room for improvement.In particular, a common issue both in normal and Visual SLAM algorithms is th data association problem. Wrong data association either disturbs stability orresult in divergence of the SLAM process. In this study, we propose two outlierelimination methods which use predicted feature location error and optical flow field.The former method asserts estimated landmark projection and its measurement locationsto be close. The latter accepts optical flow field as a reference and compares the vectoformed by consecutive matched feature locations; eliminates matches contradicting withthe local optical flow vector field. We have shown these two methods to be saving VSLAMfrom divergence and improving its overall performance. We have also described our newmodular SLAM library, SLAM++.