VLA architectures exhibit architecture-specific failure signatures at the motor-command level, with direction reversal as a universal predictor and velocity monitoring ineffective for continuous models.
Römer, A
6 Pith papers cite this work. Polarity classification is still indexing.
citation-role summary
citation-polarity summary
fields
cs.RO 6years
2026 6verdicts
UNVERDICTED 6roles
other 1polarities
unclear 1representative citing papers
A retrieve-then-steer method stores successful robot actions in memory and uses them to steer a frozen VLA's flow-matching sampler for better test-time reliability without parameter updates.
RePO-VLA raises average adversarial success rates in VLA manipulation from 20% to 75% by using recovery-aware initialization, a progress-aware semantic value function, and value-conditioned refinement on success and corrective trajectories.
Temporal difference calibration aligns uncertainty estimates in vision-language-action models with their value functions for better sequential performance.
CMP projects actions onto a learned competence manifold using a frame-wise safety scheme and isomorphic latent space to achieve up to 10x better survival in out-of-distribution scenarios with under 10% tracking loss.
ROBOGATE applies adaptive boundary-focused sampling in simulation to discover robot policy failure boundaries, revealing a 97.65 percentage point performance gap for a VLA model between LIBERO and industrial scenarios.
citing papers explorer
-
Retrieve-then-Steer: Online Success Memory for Test-Time Adaptation of Generative VLAs
A retrieve-then-steer method stores successful robot actions in memory and uses them to steer a frozen VLA's flow-matching sampler for better test-time reliability without parameter updates.