Long-WAM: Scaling the Context of World-Action Models
Long-WAM scales robot world-action context, lifting RoboCasa success from 63.3% to 78.7%.
Long-WAM scales the context of causal world-action models under real-time robot-control constraints. On RoboCasa GR-1, extending context from 0 to 19.2 seconds raises success from 63.3% to 78.7% when the video foundation is pretrained autoregressively, while a bidirectional initialization shows no net gain. It reports the best compared results on LIBERO-Long, RoboTwin 2.0, and DOMINO, with each action chunk taking 107.4 ms on an RTX 5090. Deployed on Unitree G1 and YAM, it reaches 95% success on dynamic cup stacking, where Pi0.5 and Fast-WAM succeed in none of 20 trials.
- Autoregressive pretraining makes longer visual history useful for control.
- RoboCasa GR-1 success rises from 63.3% to 78.7% with 19.2 seconds of context.
- Action chunks, including future-video prediction, take 107.4 ms on an RTX 5090.
- Dynamic cup stacking reaches 95% success versus zero for Pi0.5 and Fast-WAM.
Full article216 words · extracted from arxiv.org · click to collapse
Real-time robot control demands enough visual history to infer motion and task progress, but processing that history can delay action. We present Long-WAM, a model-system framework for scaling the context of causal world-action models under real-time control constraints. Our central finding is that access to history is not the same as using it: longer histories pay off far more when the video foundation is pretrained autoregressively (AR). We first learn causal prediction from robot and egocentric videos without action labels, then preserve this history-to-future structure during world-action adaptation. On RoboCasa GR-1, increasing context from 0.0 to 19.2 seconds raises success from 63.3% to 78.7%, whereas a bidirectionally pretrained initialization shows no net gain; robot-domain AR pretraining further raises peak success on GR-1 and LIBERO-Long. Long-WAM also achieves the best results among compared methods on LIBERO-Long, RoboTwin 2.0, and DOMINO. Streaming observation encoding, asynchronous execution, and hardware-specific acceleration enable deployment on RTX 5090, DGX Spark, and Jetson AGX Thor without dropping future prediction; on RTX 5090, each action chunk, including future-video latent prediction, takes 107.4 ms. Real-time deployment on Unitree G1 and YAM supports dynamic and long-horizon manipulation, including 95% success on dynamic cup stacking, where Pi0.5 and Fast-WAM succeed in none of 20 trials. As a memory-informed executor, Long-WAM also complements higher-level planning in composite tasks.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2610.10528