Strengthening Generative Robot Policies through Predictive World Modeling

We present generative predictive control (GPC), a learning control framework that (i) clones a generative diffusion-based policy from expert demonstrations, (ii) trains a predictive action-conditioned world model from both expert demonstrations and random explorations, and (iii) synthesizes an online planner that ranks and optimizes the action proposals from (i) by looking ahead into the future using the world model from (ii). Crucially, we show that conditional video diffusion allows learning (near) physics-accurate visual world models and enable robust visual foresight. Focusing on planar pushing with rich contact and collision, we show GPC dominates behavior cloning across state-based and vision-based, simulated and real-world experiments.

@article{qi2025_2502.00622,
  title={ Strengthening Generative Robot Policies through Predictive World Modeling },
  author={ Han Qi and Haocheng Yin and Yilun Du and Heng Yang },
  journal={arXiv preprint arXiv:2502.00622},
  year={ 2025 }
}