Simulation Distillation: Pretraining World Models in Simulation for Rapid Real-World Adaptation
This addresses the problem of sim-to-real transfer for robotics, enabling faster and more stable adaptation in low-data regimes, though it appears incremental as it builds on existing world model and planning methods.
The paper tackles the challenge of simulation-to-real transfer in robotics by introducing Simulation Distillation (SimDist), a framework that distills structural priors from a simulator into a latent world model for rapid real-world adaptation, achieving substantial improvements in data efficiency, stability, and final performance across manipulation and locomotion tasks.
Simulation-to-real transfer remains a central challenge in robotics, as mismatches between simulated and real-world dynamics often lead to failures. While reinforcement learning offers a principled mechanism for adaptation, existing sim-to-real finetuning methods struggle with exploration and long-horizon credit assignment in the low-data regimes typical of real-world robotics. We introduce Simulation Distillation (SimDist), a sim-to-real framework that distills structural priors from a simulator into a latent world model and enables rapid real-world adaptation via online planning and supervised dynamics finetuning. By transferring reward and value models directly from simulation, SimDist provides dense planning signals from raw perception without requiring value learning during deployment. As a result, real-world adaptation reduces to short-horizon system identification, avoiding long-horizon credit assignment and enabling fast, stable improvement. Across precise manipulation and quadruped locomotion tasks, SimDist substantially outperforms prior methods in data efficiency, stability, and final performance. Project website and code: https://sim-dist.github.io/