DAGS-SLAM: Dynamic-Aware 3DGS SLAM via Spatiotemporal Motion Probability and Uncertainty-Aware Scheduling

Li Zhang; Yu-An Liu; Xijia Jiang; Conghao Huang; Danyang Li; Yanyong Zhang

cs.ROFeb 25, 2026

DAGS-SLAM: Dynamic-Aware 3DGS SLAM via Spatiotemporal Motion Probability and Uncertainty-Aware Scheduling

Authors: Li Zhang, Yu-An Liu, Xijia Jiang, Conghao Huang, Danyang Li, Yanyong Zhang

Organizations: School of Mathematics, Hefei University of Technology, Hefei 230601, China · School of Software, Tsinghua University, Beijing 100084, China · School of Computer Science and Technology, University of Science and Technology of China, Hefei 230026, China

Abstract

Mobile robots and IoT devices demand real-time localization and dense reconstruction under tight compute and energy budgets. While 3D Gaussian Splatting (3DGS) enables efficient dense SLAM, dynamic objects and occlusions still degrade tracking and mapping. Existing dynamic 3DGS-SLAM often relies on heavy optical flow and per-frame segmentation, which is costly for mobile deployment and brittle under challenging illumination. We present DAGS-SLAM, a dynamic-aware 3DGS-SLAM system that maintains a spatiotemporal motion probability (MP) state per Gaussian and triggers semantics on demand via an uncertainty-aware scheduler. DAGS-SLAM fuses lightweight YOLO instance priors with geometric cues to estimate and temporally update MP, propagates MP to the front-end for dynamic-aware correspondence selection, and suppresses dynamic artifacts in the back-end via MP-guided optimization. Experiments on public dynamic RGB-D benchmarks show improved reconstruction and robust tracking while sustaining real-time throughput on a commodity GPU, demonstrating a practical speed-accuracy tradeoff with reduced semantic invocations toward mobile deployment.

Figures & tables

Panel of Figure 1 — Fig. 1: Typical failure cases of 3DGS-SLAM in dynamic scenes. Moving objects are often fused into the Gaussian map, leading to ghosting/blobs, smeared textures, and broken surfaces, which in turn degrades both tracking stability and static-scene reconstruction quality.

Fig. 2: System pipeline of DAGS-SLAM. A semantic thread maintains a temporally updated per-Gaussian MP state by fusing lightweight instance priors with geometric cues, and an uncertainty-aware scheduler triggers semantics on demand to reduce overhead. MP is propagated to the tracking thread for dynamic-aware correspondence filtering and to the mapping thread for MP-guided rendering optimization, producing clean static-scene reconstructions under dynamic interference.

Fig. 3: Illustration of the proposed labeling process, where yellow denotes dynamic outliers and green denotes static features.

Fig. 4: The uncertainty-aware semantic-on-demand scheduling mechanism. MP uncertainty and tracking/geometry inconsistency are fused into a trigger score to decide whether to refresh the instance prior. Segmentation is invoked only at trigger moments; otherwise the prior is reused and MP is updated temporally.

Fig. 5: Epipolar-verified static recovery and densification. We first obtain a coarse MP-based dynamic mask to localize ambiguous boundary regions, then use epipolar geometry to validate static correspondences and filter dynamic outliers. The verified static regions are restored and incorporated into the densification stage.

Fig. 6: MP-guided rendering and joint optimization. We convert per-Gaussian MP to a pixel-wise static confidence map to down-weight dynamic regions during rasterization. The objective combines MP-weighted photometric/depth losses, a static motion-consistency regularizer, and an MP-guided edge-warp term to improve alignment near motion boundaries under imperfect priors.

Sequence	ORB-SLAM3		Photo-SLAM		SplaTAM		DDN-SLAM		DG-SLAM		Ours
	ATE	Std.	ATE	Std.	ATE	Std.	ATE	Std.	ATE	Std.	ATE	Std.
fr3/walking/xyz	$0.3729$	$0.1966$	$0.3054$	$0.1482$	$0.4323$	$0.1381$	$0.0148$	$0.0096$	$0.0166$	$0.0088$	0.0128	0.0080
fr3/walking/xyz_val	$0.6821$	$0.5125$	$0.6147$	$0.3986$	$0.8849$	$0.8096$	0.0112	$0.0071$	$0.0215$	$0.0108$	$0.0131$	0.0054
fr3/walking/static	$0.1769$	$0.0968$	$0.1347$	$0.0118$	$0.3470$	$0.4326$	$0.0118$	$0.0060$	0.0062	$0.0101$	$0.0069$	0.0032
fr3/walking/static_val	$0.2784$	$0.0713$	$0.2634$	$0.0728$	$0.7368$	$0.4290$	$0.0138$	0.0079	$0.0208$	$0.0311$	0.0109	$0.0087$
fr3/walking/rpy	$0.6129$	$0.2696$	$0.5644$	$0.2789$	—		$0.0421$	$0.0266$	$0.0435$	0.0209	0.0318	$0.0292$

TABLE I: Absolute trajectory error (ATE, RMSE) on TUM RGB-D dynamic sequences ( fr3/walking/* ).

Sequence	ORB-SLAM3		Photo-SLAM		SplaTAM		DDN-SLAM		DG-SLAM		Ours
	ATE	Std.	ATE	Std.	ATE	Std.	ATE	Std.	ATE	Std.	ATE	Std.
crowd	$0.4300$	$0.0210$	$0.5104$	$0.0113$	$1.8286$	$1.2384$	$0.0191$	$0.0110$	$0.0211$	$0.0180$	0.0178	0.0090
crowd2	$0.9160$	$0.5577$	$0.9332$	$0.5789$	$3.4560$	$1.8706$	0.0226	0.0109	$0.0322$	$0.0245$	$0.0231$	$0.0127$
person_tracking	$0.6121$	$0.3220$	$0.3293$	$0.4518$	$0.5290$	$0.1022$	$0.0431$	0.0168	$0.0458$	$0.0340$	0.0420	$0.0179$
person_tracking2	$0.6568$	$0.5564$	$0.3245$	$0.2681$	$0.6606$	$0.4266$	0.0374	0.0128	$0.0690$	$0.0388$	$0.0400$	$0.0219$
synchronous	$0.7670$	$1.2382$	$0.4562$	$0.5641$	—		$0.0300$	$0.0289$	$0.0491$	$0.0284$	0.0280	0.0211

TABLE II: Absolute trajectory error (ATE, RMSE) on the BONN RGB-D dataset.

Panel of Figure 9 — Fig. 7: Qualitative visualization of DAGS-SLAM on representative dynamic sequences.

Panel of Figure 10 — Fig. 8: Qualitative comparison on representative dynamic sequences (rows: methods; columns: sequences).

Method	Metric	balloon	balloon2	person_tracking	person_tracking2	move_no_box2	Avg.
ESLAM	PSNR (dB) $\uparrow$	15.49	15.40	14.86	18.79	16.88	16.28
	SSIM $\uparrow$	0.644	0.662	0.818	0.647	0.795	0.713
	LPIPS $\downarrow$	0.529	0.560	0.345	0.416	0.581	0.486
RoDyn-SLAM	PSNR (dB) $\uparrow$	15.77	16.66	17.95	19.41	15.34	17.03
	SSIM $\uparrow$	0.593	0.608	0.829	0.658	0.675	0.673
	LPIPS $\downarrow$	0.396	0.419	0.326	0.339	0.500	0.396

TABLE III: View synthesis results (mapping quality) on dynamic sequences from the BONN dataset.

Method	Metric	balloon	balloon2	person_tracking	person_tracking2	move_no_box2	Avg.
Co-SLAM	Acc. [cm] $\downarrow$	16.70	31.19	35.41	54.22	17.12	30.93
	Comp. [cm] $\downarrow$	27.29	40.32	201.40	118.47	20.47	81.59
	Comp. Ratio [ $\leq$ 5cm %] $\uparrow$	3.11	3.30	1.98	2.85	24.91	7.23
ESLAM	Acc. [cm] $\downarrow$	17.22	26.78	59.20	89.34	12.27	40.96
	Comp. [cm] $\downarrow$	9.09	13.60	145.81	186.69	10.11	73.06
	Comp. Ratio [ $\leq$ 5cm %] $\uparrow$	25.73	21.86	20.49	17.29	41.38	25.35

TABLE IV: Geometric reconstruction on dynamic scenes from the BONN dataset.

Panel of Figure 13 — Fig. 9: Rendering ablation on a representative sequence; the rightmost column shows a close-up for visual inspection.

Dataset	Metric	w/o MP	w/o epi-verify	w/o backend	Ours
BONN	ATE [m] $\downarrow$	$0.5217$	$0.0451$	$0.0315$	0.0212
BONN	Std. [m] $\downarrow$	$0.3413$	$0.0276$	$0.0197$	0.0135
TUM	ATE [m] $\downarrow$	$0.3521$	$0.0322$	$0.0230$	0.0202
TUM	Std. [m] $\downarrow$	$0.2819$	$0.0198$	$0.0226$	0.0171

TABLE V: Tracking ablation on TUM and BONN.

Setting	Metric	crowd	person_tracking	move_no_box2	Avg.
w/o MP	PSNR (dB) $\uparrow$	13.63	18.36	17.32	16.43
w/o MP	SSIM $\uparrow$	0.612	0.913	0.730	0.751
w/o epi-verify	PSNR (dB) $\uparrow$	14.78	17.91	19.21	17.30
w/o epi-verify	SSIM $\uparrow$	0.519	0.816	0.657	0.664
w/o backend	PSNR (dB) $\uparrow$	17.01	18.87	19.62	18.50
w/o backend	SSIM $\uparrow$	0.718	0.829	0.837	0.794

TABLE VI: Mapping-quality ablation on BONN.

Variant	ATE $\downarrow$	Std. $\downarrow$	PSNR $\uparrow$	Flip ratio $\downarrow$
Instant MP (per-frame)	0.0286	0.0175	21.92	7.8
Temporal MP (fixed)	0.0235	0.0150	22.34	5.4
Temporal MP (adaptive)	0.0212	0.0135	22.67	3.6

TABLE VII: Ablation on temporal MP and uncertainty-aware update.

Variant	ATE $\downarrow$	Std. $\downarrow$	PSNR $\uparrow$	Flip $\downarrow$
MP ${}_{\text{ins}}$ only ( $K{=}0$ )	0.0397	0.0236	21.35	8.3
MP ${}_{\text{geo}}$ only ( $K{=}1$ )	0.0338	0.0204	20.72	7.1
Fusion	0.0289	0.0162	22.41	4.6

TABLE VIII: Ablation on MP fusion: MP

{}_{\text{ins}}

only vs. MP

{}_{\text{geo}}

only vs. fusion.

Sched.	ATE $\downarrow$	Std. $\downarrow$	PSNR $\uparrow$	SSIM $\uparrow$	Calls $\downarrow$	FPS $\uparrow$
Always	0.0210	0.0134	22.68	0.839	$420$	$25.6$
Fixed ( $N{=}10$ )	$0.0213$	$0.0136$	$22.61$	$0.837$	$180$	$36.2$
On-Demand (Ours)	$0.0212$	$0.0135$	$22.67$	0.839	140	40.5

TABLE IX: Accuracy–efficiency trade-off of semantic scheduling on the BONN subset.

Strategy	Track (ms)	Map (ms)	MP (ms)	YOLO (ms)	YOLO (%)	Total (ms)
Always	$7.6$	$13.4$	$1.2$	$16.8$	$42.9$	$39.0$
Fixed ( $N{=}10$ )	$7.6$	$13.4$	$1.2$	$5.4$	$19.7$	$27.6$
On-Demand	$7.6$	$13.4$	$1.2$	$2.5$	$10.1$	$24.7$

TABLE X: Latency breakdown and semantic overhead under different scheduling strategies (BONN subset).

Method	Map. (ms) $\downarrow$	YOLO (ms) $\downarrow$	#YOLO $\downarrow$	FPS $\uparrow$	Params (M) $\downarrow$	GPU (GB) $\downarrow$
NICE-SLAM	$291.8$	—	—	$1.2$	$11.9$	$10.9$
DN-SLAM	$61.9$	—	—	$4.2$	$30.1$	5.8
DDN-SLAM	$50.7$	—	—	$20.1$	$19.6$	$6.2$
Ours	13.4	2.5	140	40.5	10.2	$6.4$

TABLE XI: Comprehensive efficiency comparison on the BONN dataset.

Method	Corruption	ATE $\downarrow$	Std. $\downarrow$	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
DG-SLAM	None	0.0392	0.0238	18.10	0.767	0.378
DG-SLAM	Mild	0.0615	0.0369	16.92	0.728	0.441
DG-SLAM	Severe	0.0964	0.0578	14.83	0.661	0.523
DAGS-SLAM	None	0.0286	0.0157	22.73	0.855	0.252
DAGS-SLAM	Mild	0.0314	0.0179	22.10	0.846	0.268
DAGS-SLAM	Severe	0.0398	0.0241	20.82	0.822	0.310

TABLE XII: Robustness to corrupted instance priors on BONN dataset.

Dynamic SLAM

See all 12 papers

DL-SLAM: Enabling High-Fidelity Gaussian Splatting SLAM in Dynamic Environments based on Dual-Level Probability
Jul 2, 2026Ziheng Xu, Qingfeng Li, Xuefeng Liu +2Simultaneous Localization and Mapping Dynamic Scene Reconstruction
ACMMM 2026, , Rio de Janeiro, Brazil
Flow4DGS-SLAM: Optical Flow-Guided 4D Gaussian Splatting SLAM
Apr 24, 2026Yunsong Wang, Gim Hee LeeSimultaneous Localization and Mapping 3D Gaussian Splatting
School of Computing, National University of Singapore
GGD-SLAM: Monocular 3DGS SLAM Powered by Generalizable Motion Model for Dynamic Environments
Apr 14, 2026Yi Liu, Haoxuan Xu, Hongbo Duan +5Dynamic SLAM 3DGS SLAM
Shenzhen International Graduate School, Tsinghua University, Shenzhen, China. · Thrust of Robotics and Autonomous Systems, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China. · the School of Cyber Science and Technology, Sun Yat-sen University, Shenzhen, China. +1

Jul 2, 2026J/K move · Enter open · S save

DL-SLAM: Enabling High-Fidelity Gaussian Splatting SLAM in Dynamic Environments based on Dual-Level Probability

Ziheng Xu, Qingfeng Li, Xuefeng Liu, Chen Chen+1

ACMMM 2026, , Rio de Janeiro, Brazil

Simultaneous Localization and Mapping Dynamic Scene Reconstruction Gaussian Splatting

Open paper AskPDF arXiv

Recent advances in 3D Gaussian Splatting (3DGS) have enabled significant progress in dense dynamic Simultaneous Localization And Mapping (SLAM). Prevailing methods typically discard predefined dynamic objects, ignoring that transiently static objects offer valuable geometric constraints for pose estimation. A recent work attempts to leverage this potential by employing per-pixel uncertainty maps to quantify the magnitude of motion. While this approach enables transiently static objects to enhance pose estimation, it erroneously integrates these objects into the static map, resulting in persistent artifacts. Moreover, its reliance on purely geometric information leads to ambiguous object boundaries in the uncertainty maps. To overcome these limitations, we present DL-SLAM, a monocular Gaussian Splatting SLAM system built upon a novel dual-level probabilistic framework. Our method computes dynamic probability maps by combining semantic and geometric information. These pixel-level probabilities are lifted to 3D and aggregated to derive an object-level dynamic probability for each instance. Object-level probability enables the categorical pruning of dynamic Gaussians, resulting in an artifact-free static map. The static map, in turn, provides a geometrically consistent guidance to refine the pixel-wise probabilities, enhancing their reliability. Experimental results demonstrate that DL-SLAM outperforms existing approaches, improving tracking accuracy by up to 13% while generating high-fidelity semantic maps.

Abstract

Dynamic SLAM

Explore similar work

DL-SLAM: Enabling High-Fidelity Gaussian Splatting SLAM in Dynamic Environments based on Dual-Level Probability

Flow4DGS-SLAM: Optical Flow-Guided 4D Gaussian Splatting SLAM

GGD-SLAM: Monocular 3DGS SLAM Powered by Generalizable Motion Model for Dynamic Environments