goal-conditioned-control

Companion code to our paper on connections between goal-conditioned reinforcement learning and optimal control.

Set-up

Create a virtual enviroment and install dependencies:

uv init
uv sync

I couldn't install Neuromancer (for the double inverted pendulum example) via uv at this time. For that example, simply run the notebook, which will handle installation via pip.

Citation

To reference this work, please use the following bib entry:

@article{lawrence2025goalconditioned,
  title={Why Goal-Conditioned Reinforcement Learning Works: Relation to Dual Control},
  author={Lawrence, Nathan P and Mesbah, Ali},
  journal={arXiv preprint arXiv:2512.06471},
  year={2025}
}