Robust visual SLAM with point and line features

Abstract

Abstract-In this paper, we develop a robust efficient visual SLAM system that utilizes heterogeneous point and line features. By leveraging ORB-SLAM [1], the proposed system consists of stereo matching, frame tracking, local mapping, loop detection, and bundle adjustment of both point and line features. In particular, as the main theoretical contributions of this paper, we, for the first time, employ the orthonormal representation as the minimal parameterization to model line features along with point features in visual SLAM and analytically derive the Jacobians of the re-projection errors with respect to the line parameters, which significantly improves the SLAM solution. The proposed SLAM has been extensively tested in both synthetic and real-world experiments whose results demonstrate that the proposed system outperforms the state-of-the-art methods in various scenarios.

Robust RGB-D SLAM Using Point and Line Features for Low Textured Scene

“…To avoid overparameterization problem caused by the Plücker coordinate with six parameters, orthonormal representation is applied. We simply give the convention between orthonormal representation and Plücker coordinate below, and the detail can be referred to [ 16 , 36 ]. where is a 3D rotation matrix and is the rotation vector.…”

“…Despite low texture, most indoor scenes contain abundant high-level geometry primitives such as lines and planes, which can be fused to aid camera tracking. Lines have been widely applied in monocular and stereo SLAM systems [ 10 , 11 , 12 , 13 , 14 , 15 , 16 ] but attracted less attention in the RGB-D research field [ 17 , 18 , 19 ]. Moreover, the existing line-based methods exploit either 3D–2D line correspondences or 3D–3D line correspondences [ 17 , 18 , 19 ].…”

Automatic sensor orientation using horizontal and vertical line feature constraints

“…where Z denotes measured image point projections. ρ f , ρ g , ρ h and ρ v are robust Huber cost functions (Zuo et al, 2017).…”

“…In order to overcome failure of Perspective-n-Point (PnP) problem in low texture environment, (Vakhitov et al, 2016) introduced an algebraic line error formulated as linear constraints on the line endpoints so as to handle points, lines, or a combination of them. (Zuo et al, 2017) employed orthonormal representation as the minimal parameterization of modelling line features and then developed robust efficient visual SLAM system based heterogeneous point and line features, significantly improving the SLAM solution. (Hofer et al, 2017) utilized both straight line and point feature to achieve the task of 3D scene abstraction so as to overcome the high computational complexity associated with subsequent multi-view stere-o approaches.…”

PL-VIO: Tightly-Coupled Monocular Visual–Inertial Odometry Using Point and Line Features

“…Bartoli and Sturm [ 30 ] proposed the orthonormal representation, which uses a three-DoF rotation matrix and a one-DoF rotation matrix to update the line parameters during optimization. The orthonormal representation has been used in some stereo visual SLAM systems [ 27 , 31 ]. For VIO approaches, Kottas and Roumeliotis [ 26 ] investigated the observability of the VIO using line features only.…”