Vision-Language-Action (VLA) policies often rely strongly on the camera viewpoints seen during training, causing substantial performance degradation when deployed from unseen perspectives. Collecting demonstrations from sufficiently diverse physical viewpoints is expensive and still provides only sparse coverage of the viewpoint space. We introduce InfiNoVA, a data-augmentation framework that converts synchronized multi-camera demonstrations into a dense distribution of geometrically consistent training views. InfiNoVA reconstructs each manipulation trajectory as a time-varying 3D Gaussian representation and renders novel observations from sampled camera poses while preserving the original state-action correspondence. This explicit scene representation improves frame-level fidelity and temporal consistency while reducing task-critical hallucinations observed in generative novel-view synthesis. Across four real-world manipulation tasks, policies trained with InfiNoVA achieve 5.4x higher average success under unseen randomized viewpoints than both VISTA-based augmentation and the unaugmented policy. InfiNoVA further achieves 1.7x higher success than training directly on all five physical camera views. These results show that dense, geometrically grounded viewpoint augmentation provides a practical route toward camera-robust robot policies without modifying the underlying policy architecture.
Figures & tables
Fig. 2: Multi-view reconstruction and novel-view sampling. Red cameras are used for reconstruction; green cameras denote sampled novel views.
Fig. 3: Qualitative comparison of novel-view synthesis across the four manipulation tasks. Rows correspond to competing synthesis methods. Red ellipses highlight representative task-critical artifacts in VISTA.
Method
PSNR ↑
SSIM ↑
LPIPS ↓
Hunyuan 3D [ 34 ]
7.48
0.693
0.596
DepthSplat [ 35 ]
13.27
0.809
0.517
GLD [ 33 ]
15.97
0.753
0.485
VISTA [ 13 ]
15.13
0.820
0.459
InfiNoVA (Ours)
16.65
0.841
0.388
TABLE I: Quantitative comparison of novel-view rendering quality. Higher PSNR and SSIM indicate better reconstruction fidelity, while lower LPIPS indicates better perceptual similarity.
Task
GLD
VISTA
InfiNoVA
Pick-and-Place
25615
1115
318
Sweep
35119
1529
437
Sort
39026
1693
434
Stack
45056
1961
560
Average/episode
36204
1575
437
TABLE II: Novel-view synthesis time comparison (in seconds).
Method
FeaCD ↓
ReStraV- Δθ ( ∘ ) ↓
Hunyuan 3D [ 34 ]
0.9780
6.09
DepthSplat [ 35 ]
0.969
5.19
VISTA [ 13 ]
0.984
7.16
InfiNoVA (Ours)
0.915
3.49
TABLE III: Temporal consistency averaged across the four manipulation tasks. Lower values indicate better agreement with the ground-truth trajectories.
Fig. 4: Temporal consistency across three timesteps ( t90 , t494 , and t512 ) sampled from the same Pick-and-Place episode. Red ellipses highlight VISTA’s temporal inconsistencies.
Method
Pairwise LPIPS ↓
LPIPS to GT ↓
GLD [ 33 ]
0.5428
0.4810
VISTA [ 13 ]
0.2765
0.5002
InfiNoVA (Ours)
0.1406
0.466
TABLE IV: Values report average pairwise LPIPS and mean LPIPS against ground truth over 5 generations for each of 5 observations.
Fig. 5: Policy success rates across four manipulation tasks.