A Performance Analysis of Omnidirectional Vision Based Simultaneous Localization and Mapping

This paper presents a performance analysis of omnidirectional vision based Simultaneous Localization and Mapping (SLAM). In omnidirectional vision based SLAM; robots perform vision based SLAM using only monocular omnidirectional cameras. In this paper, we mainly investigate the use of an omnidirectional camera for Extended Kalman Filter (EKF) based SLAM. To evaluate the success of omnidirectional vision based SLAM, we have also conducted the same simulations using a laser range finder (LRF). Main contributions of this paper are the use of an omnidirectional camera to perform SLAM in the Unified System for Automation and Robot Simulation (USARSim) environment, which is controlled by MATLAB in our study. The results of USARSim simulations show that depending on the environmental conditions omnidirectional cameras can be used as an alternative to other range bearing sensors and stereo cameras.