GrndCtrl: Grounding World Models via Self-Supervised Reward Alignment

Haoyang He^1,2    Jay Patrikar²    Dong-Ki Kim²    Max Smith²    Daniel McGann²
Ali-akbar Agha-mohammadi²    Shayegan Omidshafiei²    Sebastian Scherer<sup>1,2

1</sup> Carnegie Mellon University    ² FieldAI

Overview

Recent advances in video world modeling have enabled large-scale generative models to simulate embodied environments with high visual fidelity, providing strong priors for prediction, planning, and control. Yet, despite their realism, these models often lack geometric grounding, limiting their use in navigation tasks that require spatial coherence and long-horizon stability.

GrndCtrl introduces Reinforcement Learning with World Grounding (RLWG), a self-supervised post-training framework that aligns pretrained world models with a physically verifiable structure through geometric and perceptual rewards. Analogous to reinforcement learning from verifiable feedback (RLVR) in language models, RLWG uses multiple rewards that measure pose cycle-consistency, depth reprojection, and temporal coherence. GrndCtrl instantiates this framework using Group Relative Policy Optimization (GRPO), yielding world models that maintain stable trajectories, consistent geometry, and reliable rollouts for embodied navigation.

Code Release

Stay tuned...

Citation

If you find our work useful, please cite our paper:

@misc{he2025grndctrlgroundingworldmodels,
      title={GrndCtrl: Grounding World Models via Self-Supervised Reward Alignment}, 
      author={Haoyang He and Jay Patrikar and Dong-Ki Kim and Max Smith and Daniel McGann 
              and Ali-akbar Agha-mohammadi and Shayegan Omidshafiei and Sebastian Scherer},
      year={2025},
      eprint={2512.01952},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2512.01952}, 
}