AI summaryⓘ
The authors found that vision-language-action models for robots fail when instructions are rephrased, not because they don't understand the task, but due to how visual and text information get mixed together. This mixing causes changes in features that confuse the robot's action decisions. They propose a new method called Grounded Semantic Re-binding (GSR) that keeps text and visuals separate and then combines them, which helps the robot understand instructions better without needing massive amounts of extra data. Their approach improved performance a lot on a standard test and led to a model that handles reworded instructions almost perfectly. Overall, the authors show that smarter design can fix this problem more efficiently than just adding more training data.
Vision-Language-Action (VLA) modelsrobotic manipulationinstruction paraphrasingfeature shiftsemantic groundingaction policyjoint encodingGrounded Semantic Re-binding (GSR)instruction robustnessLIBERO-Para benchmark
Abstract
Vision-Language-Action (VLA) models excel in robotic manipulation but suffer catastrophic performance drops when canonical instructions are simply paraphrased. Although this brittleness is typically addressed through costly data scaling, our probing reveals that the root cause is architectural rather than a lack of semantic understanding. Specifically, we demonstrate that current VLAs successfully retain the correct task identity internally. The failure actually stems from the joint encoding of dynamic visual observations and text, which introduces systematic feature shifts. Because the downstream action policy is highly vulnerable to these variations, it fails to translate the preserved semantics into correct control commands. To resolve this structural bottleneck, we propose Grounded Semantic Re-binding (GSR), an elegant intervention that bypasses unstable joint routing by explicitly fusing independently extracted task semantics with native visual features to train a completely re-initialized action expert from scratch. This targeted intervention dramatically restores paraphrastic invariance using only canonical demonstrations. On the LIBERO-Para benchmark, GSR improves success rates by up to 44.6 percent. It enables lightweight models to rival massively scaled baselines and pushes state-of-the-art models to a new record PRIDE score of 70.4, outperforming the recently introduced large-scale pretrained model Xiaomi-Robotics-0 in instruction generation capabilities. Building on these insights, we also introduce ParaVLA, a natively decoupled 0.33B-parameter model exhibiting near-perfect robustness to instruction rewording. Ultimately, our work proves that robust semantic grounding can be achieved through elegant structural design, bypassing the inefficient brute-force data scaling paradigm.