Robot-Aligned 3D Gaussian Splatting for Precise Reaching with Wrist-Camera VLA Policies

Zuoxu Wang; Xiao Liang

cs.ROOct 26, 2026

Robot-Aligned 3D Gaussian Splatting for Precise Reaching with Wrist-Camera VLA Policies

Authors: Zuoxu Wang, Xiao Liang

Organizations: Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, TX 77843 USA

Abstract

Efficient robotic data collection remains a major bottleneck for vision-language-action (VLA) policy learning, especially for tasks requiring accurate end-effector positioning and camera-action alignment. This challenge is particularly severe in electronic-waste disassembly, where robots operate in cluttered assemblies of densely packed, visually similar components. Full digital-twin simulation is difficult because many parts are thin, reflective, or mechanically complex, making accurate object-level meshes and contact models difficult to obtain. In this work, we focus on pre-contact reaching and propose a 3D Gaussian Splatting (3DGS)-based data generation pipeline for wrist-camera VLA adaptation. Our method reconstructs the real workspace with 3DGS and renders diverse wrist-camera observations from robot-aligned poses. These observations are paired with smooth, structured reaching trajectories to provide robot-aligned visual-action supervision for VLA fine-tuning. We further augment the VLA policy with an instruction-conditioned reach verifier that estimates target proximity and provides reliable termination before physical interaction. Experiments in motherboard disassembly demonstrate precise VLA-based reaching using wrist-camera observations alone, without an external side-view camera. With $π$ 0.5, 1,500 generated trajectories achieve 80.0% precise-reaching success across three targets, compared with 58.7% using 150 teleoperated demonstrations, while requiring less than 10 minutes of active human capture. Project website: https://001-wang.github.io/PrecisionReach-GS/

Figures & tables

Fig. 1: Overview of the proposed real-to-sim-to-real pipeline. The system first captures and calibrates the real workspace, then reconstructs and aligns a 3DGS scene for robot-feasible wrist-camera data generation, and finally fine-tunes and deploys a VLA policy for closed-loop precise reaching.

Method	3DGS	Physics Sim.	Object Mesh	GPU Simulator Req.
SplatSim [ 39 ]	✓	✓	$\times$	✓
RoboGSim [ 36 ]	✓	✓	✓	✓
Real2Render2Real [ 33 ]	✓	$\times$	✓	✓
PrecisionReach-GS	✓	$\times$	$\times$	$\times$

TABLE I: Comparison with representative 3DGS-based robotic data generation methods.

Panel of Figure 3 — Fig. 2: Rendering representations reconstructed from the same workspace capture. The 3DGS representation better preserves the appearance needed for wrist-camera policy observations.

Fig. 3: Overview of the instruction-conditioned reach verifier. The verifier integrates the current wrist-camera observation, language instruction, and robot state to estimate whether the end-effector has reached the target interaction region. Language-conditioned FiLM and spatial attention adapt the visual representation to the instructed target, while state masking discourages reliance on robot-state shortcuts. During deployment, reaching is confirmed only after three consecutive valid predictions.

Fig. 4: Conceptual illustration of progressive wrist-camera feedback for precise reaching. The robot first performs coarse target localization, then refines its motion through repeated visual feedback, and finally stops when the reach verifier confirms alignment with the target interaction region.

Panel of Figure 6 — Fig. 5: Qualitative progression of three real-robot reaching trajectories (top to bottom), each shown as six frames from the initial configuration to the final target interaction pose.

Fig. 6: Real-robot reaching performance on a physically repositioned motherboard workspace that differs from the scene used for 3DGS reconstruction and data generation. Detailed trial counts are reported in Table II .

Policy	Target	Metric	Teleop Trajectories			PrecisionReach-GS Trajectories
			50	100	150	300	900	1500
$\pi_{0.5}$	Screw	Coarse reach	10/25	20/25	25/25	5/25	23/25	25/25
	Screw	Precise reach	3/25	10/25	12/25	0/25	14/25	21/25
	GPU	Coarse reach	8/25	23/25	25/25	15/25	23/25	25/25
	GPU	Precise reach	4/25	11/25	14/25	0/25	14/25	19/25
	RAM	Coarse reach	10/25	22/25	25/25	16/25	24/25	25/25

TABLE II: Real-robot reaching performance under different amounts of teleoperation and PrecisionReach-GS training data. Each entry reports successful trials out of 25. Coarse reach measures arrival in the target neighborhood; precise reach additionally requires a terminal pose from which the fixed grasp-and-lift routine succeeds.

Method	# Traj.	Real Capture	Active Human Time	Total Time
Teleoperation	50	50 demos	$\sim$ 90 min	$\sim$ 90 min
	100	100 demos	$\sim$ 150 min	$\sim$ 150 min
	150	150 demos	$\sim$ 200 min	$\sim$ 200 min
PrecisionReach-GS	300	4 videos	$<5$ min	$\sim$ 45 min
	900	12 videos	$<10$ min	$\sim$ 120 min
	1500	12 videos	$<10$ min	$\sim$ 200 min

TABLE III: Data acquisition cost and trajectory smoothness for teleoperation and PrecisionReach-GS. For PrecisionReach-GS, total time is measured for a single generation window. RMS jerk is reported in

\mathrm{m/s^{3}}

; lower is better. Detailed jerk statistics are reported in Table V .

Target	Initial Visual Context	300 Traj.		1500 Traj.
		Coarse SR	Precise SR	Coarse SR	Precise SR
Screw	Full workspace context	4/10	0/10	10/10	10/10
	Limited workspace context	1/10	0/10	9/10	9/10
	No workspace context	0/10	0/10	6/10	2/10
GPU	Full workspace context	8/10	0/10	10/10	9/10
	Limited workspace context	6/10	0/10	9/10	9/10

TABLE IV: Initial-view robustness under different generated data scales and target components. Starting views are grouped by the amount of workspace context visible in the wrist-camera observation.

Method	Mean $\downarrow$	Median $\downarrow$	P95 $\downarrow$	Max $\downarrow$
Teleoperation	3.2657	3.2532	4.1228	4.5666
PrecisionReach-GS	0.2353	0.1683	0.6969	0.9853

TABLE V: Trajectory smoothness comparison between teleoperation and PrecisionReach-GS. RMS jerk is reported in

\mathrm{m/s^{3}}

; lower values indicate smoother trajectories.

Panel of Figure 12 — Fig. 7: Representative recovery trajectory under the no-workspace-context condition. The top row shows the robot pose and the bottom row shows the corresponding wrist-camera observation. Initially, the motherboard is completely outside the wrist camera’s field of view. As the robot moves, the workspace gradually becomes visible, allowing the policy to continue toward the target region.

Controlled factor	Original setting	Matched setting	Precise success (%)
Observation modality	Teleop-150, wrist+side	Teleop-150, wrist only	$58.7\rightarrow 53.3$
Trajectory timing	GS-900, smoothstep	GS-900, linear	$58.7\rightarrow 50.7$

TABLE VI: Controlled analysis with

\pi_{0.5}

(25 trials per target).

Metric	Full	No verifier	No KF weighting
Coarse success (%)	93.3	93.3	74.7
Precise success (%)	58.7	45.3	46.7

TABLE VII: Ablation of key pipeline components with 900 generated episodes.

Verifier	MAE (cm) $\downarrow$	Near-Target MAE (cm) $\downarrow$	Stop Prec. (%) $\uparrow$	Biased-State Prec. (%) $\uparrow$
Full	$0.64\pm 0.02$	$0.10\pm 0.02$	99.8	$48.1\pm 11.6$
w/o spatial attention	$0.64\pm 0.07$	$0.12\pm 0.02$	99.8	–
w/o FiLM	$0.67\pm 0.06$	$0.12\pm 0.03$	99.5	–
w/o instruction conditioning	$0.71\pm 0.05$	$0.12\pm 0.03$	99.3	$22.8\pm 16.1$
w/o state masking	$\mathbf{0.44\pm 0.01}$	$\mathbf{0.08\pm 0.01}$	$\mathbf{100.0}$	0.0

TABLE VIII: Ablation and robustness analysis of the reach verifier. Distance metrics are evaluated using the correct target instruction. Near-target MAE is computed over frames whose ground-truth target distance is below 3 cm. State robustness is evaluated by applying a

-2

cm bias to the end-effector height.

Panel of Figure 16 — Fig. 8: Comparison between real and rendered observations.

Metric	Mean	Best	Worst
SSIM	0.682	0.919	0.503
Correlation	0.843	0.964	0.750
MSE similarity	0.979	0.991	0.963

TABLE IX: Visual similarity between real and rendered robot videos.

Metric	S1	S2	S3
Position RMSE (m)	0.0114	0.0094	0.0096
Position mean error (m)	0.0102	0.0077	0.0076
Position max error (m)	0.0258	0.0326	0.0329
Anchor position mean error (m)	0.0039	0.0032	0.0040
Anchor position max error (m)	0.0047	0.0042	0.0067
Orientation mean error (deg)	0.5792	0.4894	0.4614

TABLE X: Robot-frame alignment accuracy after transforming the reconstructed scene into the robot coordinate frame.

Robotic Manipulation

Robotic manipulation studies how robots perceive, plan, and control physical interactions to grasp, move, assemble, and use objects.

See all 1,685 papers

WARP-VLA: Wrist-Camera Adaptation for View-Robust Policy Execution in Vision-Language-Action Models
Oct 8, 2026Junmyeong Lee, Dongmin Shin, Min-Gyu Park +3Robustness of VLA Models Vision-Language-Action Models
Yonsei University. · Korea Electronics Technology Institute(KETI).
VersaCamVLA: Camera-Configurable VLA Policies for Robotic Manipulation
Oct 8, 2026Boyao Han, Chen Shi, Jingjing Qian +2Vision-Language-Action Models Robotic Manipulation
The Chinese University of Hong Kong, Shenzhen · Shenzhen Loop Area Institute · Harbin Institute of Technology, Shenzhen
ActGaze: Learning Action-Grounded Gaze through Counterfactual Visual Interventions for High-Precision Manipulation
Sep 24, 2026Jinxuan Zhu, Jiaheng Wang, Chao Tang +7Visuomotor Policy Learning Robotic Manipulation
Samsung Robotics eXperience · National University of Singapore

Oct 8, 2026J/K move · Enter open · S save

WARP-VLA: Wrist-Camera Adaptation for View-Robust Policy Execution in Vision-Language-Action Models

Junmyeong Lee, Dongmin Shin, Min-Gyu Park, Wooseok Jeon+2

Yonsei University. · Korea Electronics Technology Institute(KETI).

Robustness of VLA Models Vision-Language-Action Models Test-Time Adaptation of VLA Models Robot Manipulation Benchmarks

Open paper AskPDF arXiv

Despite recent advances in Vision-Language-Action models (VLAs) for robotic manipulation, their performance remains sensitive to changes in camera configuration. The problem becomes more evident in cross-setup deployment, as reproducing the exact camera pose used for training is nearly impossible. Unlike fixed external views, wrist views are more challenging because the camera moves with the robot, causing even small mounting variations to alter fine-grained geometric cues. To address this, we propose WARP-VLA, a camera-view robust VLA for diverse wrist camera configurations. WARP-VLA adopts a Mixture-of-Experts (MoE) architecture where individual experts learn view-specific feature transformations, and a router combines them based on implicit view information. This allows the policy to be deployed without requiring camera extrinsic parameters as additional input. Through experiments on the LIBERO benchmark, WARP-VLA improves the average success rate of pi-0.5 from 39.2% to 78.3% under wrist-view perturbations. The real-robot experiments further show that the feature-level adaptation learned in simulation successfully transfers to diverse deployment settings. To facilitate reproducibility and future research, we release our wrist viewpoint robustness benchmark and a plug-and-play implementation.

Abstract

Robotic Manipulation

Explore similar work

WARP-VLA: Wrist-Camera Adaptation for View-Robust Policy Execution in Vision-Language-Action Models

VersaCamVLA: Camera-Configurable VLA Policies for Robotic Manipulation

ActGaze: Learning Action-Grounded Gaze through Counterfactual Visual Interventions for High-Precision Manipulation