Organizations: Department of Computer and Software Engineering, Polytechnique Montreal, Montreal, Canada. · Mila - Quebec AI Institute, Montreal, Canada. · National Research Council Canada, Montreal, Canada.
Geometric foundation models estimate dense geometry and camera parameters from monocular images, but reliable SLAM also requires consistent scale and accurate data association across successive reconstructions. We present DASH-SLAM, a monocular SLAM system that integrates metric depth predictions from pretrained Depth Anything 3 into a shared-keyframe pose graph without task-specific fine-tuning. DASH-SLAM derives scale corrections from depth ratios at shared anchor images and accumulates them across reconstructions, with a damping factor controlling their influence. These corrections rescale odometry translations, loop constraints, and map geometry. Experiments on TUM, Replica, and 7-Scenes show that DASH-SLAM outperforms state-of-the-art methods. We evaluate performance using both rigid and similarity alignment to distinguish metric-scale fidelity from accuracy up to scale. We further evaluate DASH-SLAM on four challenging sequences in perceptually-degraded environments from the Unified Autonomy Stack field dataset, covering tunnels, fog, and a frozen lake. On these sequences, DASH-SLAM outperforms the foundation-model SLAM state-of-the-art and maintains tracking where classical SLAM methods fail.
Figures & tables
Fig. 1: DASH-SLAM overview. The front end selects keyframes from the incoming image stream, and the inference stage constructs submaps using Depth Anything 3. The back end propagates metric scale across consecutive submaps through their shared anchor frames. The loop-closure thread detects revisited places and passes loop constraints to the back end, which optimises the global pose graph.
Fig. 2: Visualization of a robot’s trajectory and map obtained online with DASH-SLAM in a laboratory setting. Submaps are identified in different colors. In this real-world experiment, DASH-SLAM achieved 13.57 FPS and a peak RAM usage of 14.51 GB on an NVIDIA Jetson AGX Orin.
Sim(3)
Average
360
desk
desk2
floor
plant
room
rpy
teddy
xyz
Sim(3)
SE(3)
FPS
Calibrated
ORB-SLAM3 [ 2 ]
X
0.016
X
0.015
0.111
X
0.062
X
0.009
X
X
25.2
DPV-SLAM [ 14 ]
0.153
0.021
0.038
0.130
0.024
0.330
0.030
0.084
0.013
0.091
0.398
16.9
DPV-SLAM++ [ 14 ]
0.175
0.021
0.037
0.137
0.020
0.236
0.026
0.090
0.012
0.084
63.971
16.8
DROID-SLAM [ 23 ]
0.155
0.017
0.504
0.025
0.017
0.071
0.057
0.045
0.040
0.103
0.258
33.6
MASt3R-SLAM [ 17 ]
0.048
0.016
0.024
0.025
0.020
0.061
0.023
0.040
0.009
0.030
0.116
8.9
TABLE I: ATE RMSE (m) on TUM RGB-D
Sim(3)
SE(3)
FPS
Calibrated
ORB-SLAM3 [ 2 ]
0.019
0.659
26.0
DPV-SLAM [ 14 ]
0.899
8.449
15.0
DPV-SLAM++ [ 14 ]
0.935
5.865
15.6
DROID-SLAM [ 23 ]
0.681
0.806
74.9
MASt3R-SLAM [ 17 ]
0.015
0.093
13.6
Uncalibrated
DROID-SLAM [ 23 ]
0.803
1.694
75.8
TABLE II: Average ATE RMSE (m) and runtime (FPS) on Replica.
Sim(3)
SE(3)
FPS
Calibrated
ORB-SLAM3 [ 2 ]
0.130
0.487
24.9
DPV-SLAM [ 14 ]
0.622
8.482
17.8
DPV-SLAM++ [ 14 ]
0.630
4.252
17.6
DROID-SLAM [ 23 ]
0.341
0.433
66.6
MASt3R-SLAM [ 17 ]
0.049
0.134
9.5
Uncalibrated
DROID-SLAM [ 23 ]
0.336
0.447
67.3
TABLE III: Average ATE RMSE (m) and runtime (FPS) on 7-Scenes.
DASH-SLAM
DA3-Streaming
Dataset
Scale
Deviation
Scale
Deviation
TUM
1.021
2.1%
0.871
12.9%
Replica
1.050
5.0%
1.041
4.1%
7-Scenes
1.008
0.8%
0.971
2.9%
TABLE IV: Estimated trajectory scale and deviation from metric scale.
Fyllin.
Hornb.
Frozen
Campus
lake
fog
ORB-SLAM3 [ 2 ]
×
×
6.84
1.68
VGGT-SLAM2.0 [ 15 ]
196.93
148.90
76.06
46.08
DASH-SLAM
64.19
75.84
74.35
2.66
TABLE V: UAS ATE RMSE (m) with Sim(3) alignment
Fig. 3: Ground truth and estimated UAS trajectories after Sim(3) and SE(3) alignment. Crosses mark the last ORB-SLAM3 poses before tracking failures.
Cameras are ubiquitous sensors in robotics due to their compact form factor and the perceptual richness captured through visual information. Monocular SLAM enables robots to understand the environment with a minimum setup, however, it inherently suffers from scale ambiguity. A common solution is to provide multi-modal sensor configurations, such as visual-inertial systems, where scale is observable unless the robot navigates under a constant-velocity motion, a common scenario in mobile robotics. With the advent of deep-learning, geometric foundation models have been used to address this problem, but the depths maps are often noisy and scale-inconsistent across frames. In this paper, we propose Scalix, a real-time monocular SLAM framework that achieves metric-scale state estimation by integrating learned depth cues into a probabilistic factor-graph formulation. By augmenting existing monocular depth models with both per-pixel depth uncertainty and per-frame scale uncertainty, Scalix treats scale predictions as independent measurements within its optimization, leading to improved scale consistency through multi-view data associations. Experiments in large-scale outdoor and indoor environments demonstrate state-of-the-art performance on both metric and up-to-scale benchmarks while maintaining real-time operation and generalization.
Sebastian Barbas Laina, Tianyi Zhang, Panagiotis Petropoulakis +5
Monocular SLAM historically suffers from scale ambiguity and tracking failure in dynamic environments. While recent vision foundation models (VFMs) provide remarkable zero-shot depth priors, naively integrating these deterministic predictions ignores predictive uncertainty and frame-to-frame scale inconsistencies. We propose PRISM-SLAM, a real-time framework that rigorously integrates VFM priors into a structured Bayesian factor graph to achieve scale-aware, metric-consistent localization and mapping. Specifically, we introduce a Plücker Ray-Distance Factor to anchor monocular observations in absolute space within a globally consistent metric coordinate system, mathematically resolving scale drift by making the metric scale Fisher-identifiable. To handle environmental dynamics, we derive an epistemic uncertainty proxy from temporal depth consistency and formulate a Dynamic Scene Uncertainty Gating (DSUG) mechanism. This soft-gating approach probabilistically down-weights dynamic distractors without incurring the heavy computational overhead associated with traditional semantic segmentation masks. By employing a multi-process architecture that asynchronously processes VFM inference and geometric tracking, PRISM-SLAM provides verified metric output at 30 FPS using solely RGB input, bridging the gap between foundation models and real-world robotic applications. Evaluated on the TUM RGB-D and 7-Scenes benchmarks, PRISM-SLAM achieves a metric SE(3) Absolute Trajectory Error (ATE) nearly identical to its oracle-aligned Sim(3) error. This demonstrates that our system can produce deployment-ready metric trajectories by delivering robust metric SLAM solutions without any post-hoc scale correction. Project page: https://prismslam-cmd.github.io/prismslam_pr/
Recent advances in geometric foundation models have emerged as a promising alternative for addressing the challenge of dense reconstruction in monocular visual simultaneous localization and mapping (SLAM). Although geometric foundation models enable SLAM to leverage variable input views, the previous methods remain confined to two-view pairs or fixed-length inputs without sufficient deliberation of geometric context for view selection. To tackle this problem, we propose AIM-SLAM, a dense monocular SLAM framework that exploits an adaptive and informative multi-view keyframe prioritization with dense pointmap predictions from visual geometry grounded transformer (VGGT). Specifically, we introduce the selective information- and geometric-aware multi-view adaptation (SIGMA) module, which employs voxel overlap and information gain to retrieve a candidate set of keyframes and adaptively determine its size. Furthermore, we formulate a joint multi-view Sim(3) optimization that enforces consistent alignment across selected views, substantially improving pose estimation accuracy. The effectiveness of AIM-SLAM is demonstrated on real-world datasets, where it achieves state-of-the-art pose estimation performance and accurate dense reconstruction results. Our system supports ROS integration, with code is available at https://aimslam.github.io/.
Jinwoo Jeon, Dong-Uk Seo, Eungchang Mason Lee +1
The School of Electrical Engineering, KAIST (Korea Advanced Institute of Science and Technology), Daejeon, 34141, Republic of Korea · KAIST InnoCORE LLM, KAIST, Daejeon, 34141, Republic of Korea