Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

SCALE: Self-uncertainty Conditioned Adaptive Looking and Execution for Vision-Language-Action Models

About

Vision-Language-Action (VLA) models have emerged as a promising paradigm for general-purpose robotic control, with test-time scaling (TTS) gaining attention to enhance robustness beyond training. However, existing TTS methods for VLAs require additional training, verifiers, and multiple forward passes, making them impractical for deployment. Moreover, they intervene only at action decoding while keeping visual representations fixed-insufficient under perceptual ambiguity, where reconsidering how to perceive is as important as deciding what to do. To address these limitations, we propose SCALE, a simple inference strategy that jointly modulates visual perception and action based on 'self-uncertainty', inspired by uncertainty-driven exploration in Active Inference theory-requiring no additional training, no verifier, and only a single forward pass. SCALE broadens exploration in both perception and action under high uncertainty, while focusing on exploitation when confident-enabling adaptive execution across varying conditions. Experiments on simulated and real-world benchmarks demonstrate that SCALE improves state-of-the-art VLAs and outperforms existing TTS methods while maintaining single-pass efficiency.

Hyeonbeom Choi, Daechul Ahn, Youhan Lee, Taewook Kang, Seongwon Cho, Jonghyun Choi• 2026

Related benchmarks

TaskDatasetResultRank
Robot ManipulationLIBERO
Object Achievement98.7
1025
Robotic ManipulationLIBERO Spatial Object Goal Long
Overall Success Rate (Long)63.3
91
Pick-&-PlaceSIMPLER-WidowX
Average Success Rate49
21
Multi-task Robot ManipulationLIBERO-PRO-Long unseen benchmark
Language Error51.2
10
Pick-&-PlaceReal-world 'Put A on B' pick-and-place (In-Distribution)
SR (Carrot/Towel)87.5
4
Pick-&-PlaceReal-world 'Put A on B' pick-and-place Out-of-Distribution
SR (Teddy Bear on Bowl)50
4
Robotic ManipulationLIBERO Small Viewpoint Shift
Success Rate0.85
2
Robotic ManipulationLIBERO Medium Viewpoint Shift
Success Rate71.6
2
Robotic ManipulationLIBERO Large Viewpoint Shift
Success Rate62.5
2
Robotic ManipulationLIBERO Average Viewpoint Shift
Success Rate73
2
Showing 10 of 10 rows

Other info

Follow for update