Guy Le Besnerais

Many applications of Visual SLAM, such as augmented reality, virtual reality, robotics or autonomous driving, require versatile, robust and precise solutions, most often with real-time capability. In this work, we describe OV ² SLAM, a fully online algorithm, handling both monocular and stereo camera setups, various map scales and frame-rates ranging from a few Hertz up to several hundreds. It combines numerous recent contributions in visual localization within an efficient multi-threaded architecture. Extensive comparisons with competing algorithms shows the state-of-the-art accuracy and real-time performance of the resulting algorithm. For the benefit of the community, we release the source code: https://github.com/ov2slam/ov2slam.

In this paper we present a performance model for depth estimation using single image depth from defocus (SIDFD). Our model is based on an original expression of the Cramér–Rao bound (CRB) in this context. We show that this model is consistent with the expected behavior of SIDFD. We then study the influence on the performance of the optical parameters of a conventional camera such as the focal length, the aperture, and the position of the in-focus plane (IFP). We derive an approximate analytical expression of the CRB away from the IFP, and we propose an interpretation of the SIDFD performance in this domain. Finally, we illustrate the predictive capacity of our performance model on experimental data comparing several settings of a consumer camera.