Based on the aggregated data, the scale is finally recovered by solving a least-squares problem using a RANSAC-based optimizer. The framework includes a ground point extraction algorithm for selecting high-quality points on the ground plane, and a ground point aggregation algorithm for joining the extracted ground points in a local sliding window.

In this paper, with the assumption of a constant height of the camera above the ground, we develop a light-weight scale recovery framework leveraging an accurate and robust estimation of the ground plane. For applications like self-driving cars, loop closure is not always available, hence mining prior knowledge from the environment becomes a more promising approach. Typical solutions include loop closure detection and environment information mining. Scale ambiguity is a fundamental problem in monocular visual odometry. We demonstrate our system running with ORB-SLAM3 and show accurate dense depth estimation which could enable applications such as robotics and augmented reality. Our dense mapper can be used not only for local mapping but also globally consistent dense 3D reconstruction through TSDF fusion. This flexible design allows for integration with arbitrary metric sparse SLAM systems without delaying the main SLAM process. Our mapper runs in a separate thread in parallel to the SLAM system in a loosely coupled manner. The use of a VAE then enables us to refine the dense depth images through multi-view optimization which improves the consistency of overlapping frames. We build on CodeSLAM and use a variational autoencoder (VAE) which is conditioned on intensity, sparse depth and reprojection error images from sparse SLAM to predict an uncertainty-aware dense depth map. In this paper we propose a dense mapping framework to complement sparse visual SLAM systems which takes as input the camera poses, keyframes and sparse points produced by the SLAM system and predicts a dense depth image for every keyframe. While these sparse maps are useful for localization, they cannot be used for other tasks such as obstacle avoidance or scene understanding. State-of-the-art sparse visual SLAM systems provide accurate and reliable estimates of the camera trajectory and locations of landmarks. We propose a novel dense mapping framework for sparse visual SLAM systems which leverages a compact scene representation. As for planar case, our method achieves comparable accuracy to the existing method, while the time consumption is much lower. For nonplanar case, the proposed method gives improved accuracy after employing constraint weighting. Experiments are performed in both simulated and real environments. An error contribution criterion of sonar measurement is proposed to weight the constraints, which further reduces the influence of noisy input. Different from the existing method which assumes the features being on the same plane in order to meet the pre-condition of quasi-homography transformation, our method allows the observation points to be distributed spatially (not limited to a plane), thus extending the proposed method to more general application environments. Each measurement value is formulated as geometrical constraint and the recovered scale can be expressed in closed-form. In this paper, the scale ambiguity in underwater monocular visual odometry is solved with 2-D sonar measurements obtained from three viewpoints.

Plane geometry