Publikation
Self-Improving VLA Policies: Selected Diffusion Noise for Spurious-Robust Action Smoothing
Duc Minh Nguyen; Bao-Ngoc Dao; Tung M. Luu; Binh Gia Nguyen; Vinh Tong; Lianjia Liu; Vu N. Duong; Dung D. Le; Daniel Sonntag; Trung Le; Hong Anh Le; Jan Peters; An Thai Le; Minh Nhat Vu; Mathias Niepert; Khoa D. Doan; Duy M. H. Nguyen; Ngo Anh Vien
In: Computing Research Repository eprint Journal (CoRR), Vol. abs/2606.14084, Pages 1-19, arXiv, 2026.
Zusammenfassung
Diffusion-based Vision-Language-Action (VLA) policies enable strong generaliza-
tion in robotic manipulation, but remain sensitive to spurious visual correlations
and noisy action generation, leading to brittle behavior under perturbations. We
introduce SELECTED DIFFUSION NOISE (SDN), a simple, training-free test-time
method that improves both robustness and success rate by leveraging the diffusion
noise space as a controllable degree of freedom. SDN dynamically samples noise
vectors that are maximally separated from a reference set to mitigate reliance on
spurious cues, while selecting candidates that yield more coherent action trajec-
tories. This dual objective encourages stable behavior even under object-masked
observations and reduces action jitter without modifying model parameters. We
evaluate SDN on two simulation benchmarks (Google Robot, Widow-X) and two
real-world robotic datasets across multiple VLA policies, including π0, Groot-N1.5,
and Groot-N1.6. SDN consistently improves success rates by +8% in simulation
and +10% in real-world settings, while producing smoother and more stable actions.
Our results highlight that diffusion noise selection can play as an effective and
general mechanism for enhancing VLA policies at test time.
