@misc{indiciae197c28f43d14,
  title = {Breaking the Vision-Action Shortcut: Latent Interface Training for Generalizable Robotics Foundation Models},
  author = {Jianman Lin and Shailesh Shailesh and Zhongyi Luo and Jiafei Duan},
  year = {2026},
  url = {https://arxiv.org/abs/2609.12641},
  note = {Source identifier: 2609.12641}
}