Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
SimForcing addresses the challenge of learning action-conditioned robot world models by leveraging simulation for motion priors while mitigating appearance discrepancies. The framework employs latent-motion distillation to transfer motion knowledge and multi-block simulation conditioning with dropout to exploit simulation predictions without over-reliance. This approach enables the jointly trained model to generate realistic real-domain videos and improves downstream policy learning, achieving state-of-the-art performance on the Bridge dataset.
Simulation-based motion priors can be effectively distilled into real-world robot video generation, overcoming appearance gaps and boosting downstream policy performance.
Action-conditioned robot world models must respond precisely to robot trajectories while preserving realistic visual dynamics, yet learning both from heterogeneous robot videos remains challenging. Simulation offers structured motion supervision, but appearance differences hinder direct transfer, and inaccurate simulation predictions can misguide real-video generation. We present SimForcing, a simulation-guided framework that uses simulation both as a source of transferable motion knowledge and as a controllable reference for prediction. First, we transfer motion knowledge from a simulation teacher through latent-motion distillation, aligning temporal changes in latent space to internalize motion priors while mitigating the influence of appearance differences. Second, we introduce multi-block simulation conditioning with condition dropout to exploit predicted simulation trajectories without relying excessively on their accuracy. Our simulation-conditioning classifier-free guidance scheme unifies these two ideas by balancing predictions based on internalized motion knowledge with those additionally guided by simulation latents. The jointly trained student generates both simulation conditions and real-domain videos, requiring no additional world model at inference. On Bridge, SimForcing achieves the best PSNR, SSIM, LPIPS, and FVD among the compared methods without external embodied pretraining. Evaluation on InternData-A1 further supports its applicability across robot datasets. Moreover, using our trained world model to initialize a vision-language-action model improves LIBERO success, suggesting its utility for downstream policy learning. \url{https://github.com/Wang-Xiaodong1899/SimForcing}