Autonomous mobile robots in warehouses and production facilities operate in precisely the environments that defeat point-feature visual odometry: featureless aisles, uniform walls, repetitive racking and specular floors. Such environments are, however, rich in straight-edge structure: rack uprights, door frames, wall junctions and cabinet outlines present high-contrast linear edges stable across wide viewpoint changes. A line-feature extension to cuVSLAM, a CUDA-accelerated visual-SLAM system, is presented that recovers this signal. Lines are parameterised as Plücker coordinates for geometric transport and as the minimal 4-DOF orthonormal representation of Bartoli and Sturm for optimisation; a Cauchy-robust point-to-line residual with analytic Jacobians is folded directly into cuVSLAM's per-frame Gauss-Newton pose solve, and the same line state is jointly refined in the Schur-complement sliding-window bundle adjustment. An adaptive information weight causes lines to assume more of the estimation burden exactly as point features become scarce, and the extension adds residual blocks to existing solvers rather than introducing a second estimator, reducing exactly to the unmodified point pipeline when disabled. Evaluated in RGB-D mode under SE(3) Umeyama alignment with no scale correction, line features reduce mean absolute trajectory error by 38.5% over the 22 sequences of the TUM RGB-D Freiburg3 collection and by 30.1% over the 8 Replica room and office sequences, where median translation error falls by 38.9% and median rotation error by 66.4%, with zero tracking failures in any configuration tested.
Figures & tables
Fig. 1: Line-augmented cuVSLAM pipeline. The point path is unmodified; the line path runs in parallel and converges on the same Gauss–Newton pose solve, where λ(Np) of equation 9 shifts authority towards lines as point observations become scarce, and on the same bundle adjustment, where the minimal orthonormal parameterisation lets line landmarks marginalise by the same Schur complement already used for points.
Fig. 2: TUM RGB-D large_cabinet_validation: top-down trajectory (left) and APE translation over time (right). Points-only (red) accumulates drift to 0.18 m RMSE, while points+lines (blue) holds to 0.04 m by using cabinet edge lines as persistent drift anchors.
Fig. 3: LSD line detections (cyan) and point features (green) on TUM RGB-D frames. Left, large_cabinet_validation: the flat, low-texture cabinet interior starves point detection while shelf and frame edges still yield long, well-localised lines. Right, long_office_household: desk and monitor edges give clean lines alongside dense point coverage, the case where both modalities already agree.
Scene
Points-only
Points+Lines
Δ
cabinet
0.2673
0.1325
−50.4%
large_cabinet
0.2592
0.0538
−79.2%
large_cabinet_validation
0.1793
0.0408
−77.3%
long_office_household
0.0767
0.0265
−65.5%
nostructure_notexture_far
0.4358
0.2368
−45.7%
nostructure_notexture_near_withloop
0.3889
0.7386
+89.9%
TABLE I: TUM RGB-D, all 22 sequences. APE translation RMSE (m), points-only versus points+lines. Better configuration per sequence in bold.
Method
RMSE APE (m)
ORB-SLAM3
0.067
DPVO
0.80
cuVSLAM Odom (paper baseline)
0.11
cuVSLAM Points-only (our eval)
0.1112
cuVSLAM Points + LSD+LBD
0.0357
Improvement
−67.9%
TABLE II: TUM RGB-D, ten-sequence subset means. APE translation RMSE (m) for published methods and for the two configurations.
Fig. 4: LSD line detections (cyan) and point features (green) on Replica frames. Left, room0: door and window-shelf frames and a wall-mounted grid provide stable line evidence over otherwise blank wall and floor. Right, office3: wall edges and furniture outlines anchor lateral drift where the wall carries almost no point texture.
Scene
Trans (m)
Rot (deg)
Improvement
Pts
Pts+Lines
Pts
Pts+Lines
Trans
Rot
room0
0.0129
0.0079
0.239
0.163
−38.8%
−31.8%
room1
0.0124
0.0045
0.822
0.176
−63.7%
−78.6%
room2
0.0093
0.0101
0.453
0.196
+8.6%
−56.7%
office0
0.0173
0.0169
0.652
1.242
−2.3%
+90.5%
office1
0.0103
0.0091
0.997
0.547
−11.7%
−45.1%
TABLE III: Replica dataset, all 8 sequences. APE translation and rotation RMSE, points-only versus points+lines. Better configuration per sequence in bold.
This work presents a comprehensive benchmark evaluation of visual odometry (VO) and visual SLAM (VSLAM) systems for mobile robot navigation in real-world logistical environments. We compare multiple visual odometry approaches across controlled trajectories covering translational, rotational, and mixed motion patterns, as well as a large-scale production facility dataset spanning approximately 1.7 km. Performance is evaluated using Absolute Pose Error (APE) against ground truth from a Vicon motion capture system and a LiDAR-based SLAM reference. Our results show that a hybrid stack combining the cuVSLAM front-end with a custom SLAM back-end achieves the strongest mapping accuracy, motivating a deeper integration of cuVSLAM as the core VO component in our robotics stack. We further validate this integration by deploying and testing the cuVSLAM-based VO stack on an NVIDIA Jetson platform.
Charbel Abi Hana, Kameel Amareen, Mohamad Mostafa +5
Robust stereo visual-inertial odometry (VIO) remains challenging in low-texture scenes and under abrupt illumination changes, where point features become sparse and unstable, leading to ambiguous association and under-constrained estimation. Line structures offer complementary geometric cues, yet many efficient point-line systems still rely on point-guided line association, which can break down when point support is weak and may lead to biased constraints. We present a stereo point-line VIO system in which line segments are equipped with dedicated deep descriptors and matched using an entropy-regularized optimal transport formulation, which performs global mass-transport assignment and supports unmatched observations under ambiguity, outliers, and partial observations. The proposed descriptor is training-free and is computed by sampling and pooling network feature maps. To improve estimation stability, we analyze the impact of line measurement noise and introduce reliability-adaptive weighting to regulate the influence of line constraints during optimization. Experiments on EuRoC and UMA-VI, together with real-world deployments in low-texture and illumination-challenging environments, demonstrate improved accuracy and robustness over representative baselines while maintaining real-time performance.
Zikun Chen, Wentao Zhao, Yihe Niu +2
School of Automation and Intelligent Sensing, Institute of Medical Robotics, Shanghai Jiao Tong University, Shanghai 200240, China. · School of Mathematical Sciences Shanghai Jiao Tong University.
SLAM (Simultaneous Localization and Mapping) and Odometry are important systems for estimating the position of mobile devices, such as robots and cars, utilizing one or more sensors. Particularly in camera-based SLAM or Odometry, effectively tracking visual features is important as it significantly impacts system performance. In this paper, we propose a method that leverages deep learning to robustly track visual features in monocular camera images. This method operates reliably even in textureless environments and situations with rapid lighting changes. Additionally, we evaluate the performance of our proposed method by integrating it into VINS-Fusion (Monocular-Inertial), a commonly used Visual-Inertial Odometry (VIO) system.
Sanghyun Park, Soohee Han
Department of Convergence IT Engineering, POSTECH, Pohang, 37673, Korea · Department of Electrical Engineering and of Convergence IT Engineering, POSTECH, Pohang, 37673, Korea