IntentVLA: Short-Horizon Intent Modeling for Aliased Robot Manipulation
For robot learning practitioners, IntentVLA solves the aliasing problem in multimodal imitation data, improving execution stability in manipulation tasks.
IntentVLA addresses inter-chunk conflict in robot manipulation under partial observability by encoding recent visual observations into a short-horizon intent representation. It achieves improved rollout stability and outperforms strong VLA baselines across AliasBench, SimplerEnv, LIBERO, and RoboCasa.
Robot imitation data are often multimodal: similar visual-language observations may be followed by different action chunks because human demonstrators act with different short-horizon intents, task phases, or recent context. Existing frame-conditioned VLA policies infer each chunk from the current observation and instruction alone, so under partial observability they may resample different intents across adjacent replanning steps, leading to inter-chunk conflict and unstable execution. We introduce IntentVLA, a history-conditioned VLA framework that encodes recent visual observations into a compact short-horizon intent representation and uses it to condition chunk generation. We further introduce AliasBench, a 12-task ambiguity-aware benchmark on RoboTwin2 with matched training data and evaluation environments that isolate short-horizon observation aliasing. Across AliasBench, SimplerEnv, LIBERO, and RoboCasa, IntentVLA improves rollout stability and outperforms strong VLA baselines