DevGRU: Depth-guided Visual Navigation using a Collision-aware Recurrent Model

Kyung Min Han; Eunsom Kim; Young J. Kim

cs.ROAug 19, 2026

DevGRU: Depth-guided Visual Navigation using a Collision-aware Recurrent Model

Authors: Kyung Min Han, Eunsom Kim, Young J. Kim

Abstract

Existing visual navigation models often aim to develop foundation models that can generalize robot navigation across diverse platforms. However, many of these models are prone to collisions when deployed in complex indoor environments, particularly in structured layouts and narrow passages. To address this problem, we propose a depth image- and point-goal-conditioned navigation system, DevGRU. The proposed system employs an action predictor (AP) that generates collision-aware future trajectories, enabling effective avoidance of immediate obstacles. In conjunction with a collision predictor, the AP further compensates for errors accumulated in the goal pose estimation and proactively mitigates future deviations. To evaluate our method, we conducted experiments across nine different scenes and three state-of-the-art approaches - ViNT, NoMaD, and NavDP - as well as four additional variants of ViNT and NoMaD. In terms of navigation performance, DevGRU significantly outperforms ViNT and NoMaD by a large margin. In addition, the proposed model has a relatively small number of trainable parameters, resulting in the fastest inference time among the baselines, particularly outperforming NavDP by 7x in model size and 17x in inference time.

Figures & tables

Fig. 1 : A mobile robot $\mathcal{R}$ moves from $v_{i-1}$ to $v_{i}$ in a topological map $\mathcal{T}$ following $\mathcal{W}$ toward $\tilde{v}_{i}$ generated by the action predictor $\pi$ with pose and depth observations $\{(\mathbf{p},\mathbf{D})^{*}\}$ . Since imminent collision is predicted by the collision predictor $\phi$ based on the current $\mathbf{D}$ , $\tilde{v}_{i}$ is pose-corrected to $\hat{v}_{i}$ by $\pi$ . Future SG estimates such as $\tilde{v}_{i+1}$ are also adjusted.

Fig. 2 : DevGRU Architecture and Navigation Pipeline.

Panel of Figure 3 — Fig. 3 : Example RGB images (top row) and corresponding depth images (bottom row) collected from various training sites that are distinct from the evaluation environments.

Fig. 4 : Collision dataset configuration. The topological map is superimposed on top of the occupancy grid map.

Fig. 5 : The robot and its driver PC used for the experiments.

Panel of Figure 6 — Fig. 6 : Topological maps used in the experiments. From (a) to (i), the path lengths from the start to the goal are 37.1, 54.4, 56.8, 50.4, 65.3, 54.06, 12.14, 18.69, 25.77 in meters, respectively.

Panel of Figure 7 — Fig. 7 : Qualitative results on the T-B6 scene. Only our method, DevGRU, successfully completed the full navigation.

Method	T-B1		T-B2		T-B3		T-B4		T-B5		T-B6		T-B7		T-B8		T-B9		Average
Method	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$	NC $\uparrow$	SPL $\uparrow$
NoMaD	0.92	0	0.19	0	0.14	0	0.20	0	0.07	0	0.48	0	0.15	0	0.34	0	0.22	0	0.3	0
ViNT	0.86	0	0.48	0	0.52	0	0.63	0	0.05	0	0.49	0	0.91	0.44	1	0.92	0.62	0	0.62	0.15
NavDP	0.91	0.50	0.96	0.94	0.76	0.50	1	1	0.82	1	0.54	0	0.95	1	0.47	0.42	0.94	0	0.82	0.60
ViNT-FT-Full	1	0.98	0.47	0	0.26	0	0.25	0	0.04	0	0.49	0	0.49	0	0.14	0	0.28	0	0.38	0.11
ViNT-FT-FrozEnc	1	0.99	0.39	0	0.22	0	0.24	0	0.05	0	0.49	0	0.15	0	0.12	0	0.22	0	0.32	0.11

TABLE I : Navigation performance across nine benchmark scenes and the overall average.

Method	Input modality	Goal modality
NoMaD	RGB	RGB
ViNT	RGB	RGB
NavDP	RGB + Depth	Point Goal (PG)
DevGRU	Depth + Egomotion	Depth + PG

TABLE II: Input and goal modalities of each method

Model	Parameters (M) $\downarrow$	Mean Inf. Time (ms) $\downarrow$
NoMaD	19.05	123.46
ViNT	29.67	176.57
NavDP	135.72	860.18
DevGRU	19.44	50.01

TABLE III : Model efficiency comparison in terms of parameter size and inference time.

Panel of Figure 11 — Fig. 8 : The two topological maps used for the ablation study: (Top) T-A1 and (bottom) T-A2. The topological maps are shown together with occupancy maps for visualization and quantitative analysis.

Model	NC $\uparrow$	SPL $\uparrow$
DevGRU-Base	0.82	0.25
DevGRU-No-SG-Correction	0.73	0.49
DevGRU-Full	0.99	0.73

TABLE IV : Ablation study results. Higher is better for Node Coverage (NC) and SPL.

Fig. 9 : Navigation results of the three model variants on T-A2. Only DevGRU-Full successfully completed the trajectory.

Fig. 10 : The collision avoidance instance using the full DevGRU. The leftmost column shows the future SGs $\{\tilde{v}_{i},\ldots,\tilde{v}_{i+3}\}$ in red and the collision-aware waypoints $\hat{\mathcal{W}_{t}}$ in cyan w.r.t. the robot pose $\mathbf{p}_{t}$ . The center column shows an RGB–depth image pair overlaid with $\hat{\mathcal{W}_{t}}$ and $\tilde{v}_{i}$ . The rightmost column shows the RGB–depth image pair corresponding to $v_{i}$ .

Gated Recurrent Unit

See all 17 papers

Learning to Navigate with Minimal Parameters: Decomposing Visual Navigation Through Closed-Form Geometric Interfaces
Jul 13, 2026Edward Beng Wai Tan, Siew-Kei LamVisual Navigation Robot Policy Learning
College of Computing and Data Science, Nanyang Technological University, Singapore.
SAFER-Nav: Enhancing Safety for Visual Robot Navigation via Segmentation-Aware Fine-Tuning
Jun 10, 2026Geonyeong Ko, Giung Lee, Changjoo NamRobot Policy Learning Trajectory Generation
Dept. of Electronic Engineering, Sogang University, Seoul, Korea · Dept. of Computer Science, Rice University, Houston, TX, USA · Vertical Labs, Co., Ltd., Seoul, Korea
SaferPath: Hierarchical Visual Navigation with Learned Guidance and Safety-Constrained Control
Mar 2, 2026Lingjie Zhang, Zeyu Jiang, Changhao ChenVisual Robot Navigation Robot Navigation
PEAK-Lab, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, 511453, China

Jul 13, 2026J/K move · Enter open · S save

Learning to Navigate with Minimal Parameters: Decomposing Visual Navigation Through Closed-Form Geometric Interfaces

Edward Beng Wai Tan, Siew-Kei Lam

College of Computing and Data Science, Nanyang Technological University, Singapore.

Visual Navigation Robot Policy Learning Trajectory Generation Diffusion-Based Planning Robot Navigation Data-Efficient Robot Learning+1

Open paper AskPDF arXiv

Visual navigation policies have grown to hundreds of millions of parameters trained on billions of frames, with geometry, mapping, and control learned implicitly. We propose a decomposed point-goal navigation system in which operations with known closed-form structure, such as projective geometry, occupancy, and coordinate transforms, are computed analytically and serve as interfaces between three small learned modules: an egress predictor that grounds the episode goal as a local subgoal in the current view, a navigation predictor that estimates a goal-conditioned posterior over where trajectories travel, and an endpoint-pinned residual diffusion generator that samples trajectory shapes from this posterior. Only 0.58M out of 23M parameters are trained, on 44k frames, in under one GPU-hour. Across 6060 point-goal episodes in 60 environments, the system attains competitive success rates with the lowest collision rate among evaluated methods. We further show that under this decomposition, the frozen image encoder can be replaced by a 0.54M MobileNetV2 at a -2.0 SR cost, bringing the full system under 1.2M parameters. It also transfers to no-goal exploration by retraining only the 123k-parameter egress head, and its failure modes under sensor corruption are transparent and analytically correctable. We deploy and evaluate the system zero-shot on a low-cost UGV, running navigation and localization on a Jetson Orin Nano in real-time.

Abstract

Gated Recurrent Unit

Explore similar work

Learning to Navigate with Minimal Parameters: Decomposing Visual Navigation Through Closed-Form Geometric Interfaces

SAFER-Nav: Enhancing Safety for Visual Robot Navigation via Segmentation-Aware Fine-Tuning

SaferPath: Hierarchical Visual Navigation with Learned Guidance and Safety-Constrained Control