Complementing images with inertial measurements has become one of the mostpopular approaches to achieve highly accurate and robust real-time camera posetracking. In this paper, we present a keyframe-based approach tovisual-inertial simultaneous localization and mapping (SLAM) for monocular andstereo cameras. Our visual-inertial SLAM system is based on a real-time capablevisual-inertial odometry method that provides locally consistent trajectory andmap estimates. We achieve global consistency in the estimate through onlineloop-closing and non-linear optimization. Furthermore, our system supportsrelocalization in a map that has been previously obtained and allows forcontinued SLAM operation. We evaluate our approach in terms of accuracy,relocalization capability and run-time efficiency on public indoor benchmarkdatasets and on newly recorded outdoor sequences. We demonstratestate-of-the-art performance of our system compared to a visual-inertialodometry method and baseline visual SLAM approaches in recovering thetrajectory of the camera.
展开▼