MOPS🐶

Open-source code-release for paper "Meta-Optimization and Program Search using Language Models for Task and Motion Planning".

This code is build on "PRoC3S".

Setup

conda create -n "mops" python=3.10
conda activate mops
python -m pip install -e .

Add your OpenAI Key

echo "OPENAI_KEY='<YOUR-KEY-HERE>'" > .env

Example commands

The main run file is eval_policy.py. Running a particular domain involves simply creating a config file in the mops/config directory and running eval_policy.py using the --config-dir . and --config_name flags.

Here are a few example commands to give you an idea:

# Our approach on a task with goal "draw a rectangle that encloses two obstacles".
python eval_policy.py --config-name=mops_draw_pentagon

Citation

If you find our code useful for your research, please cite:

@inproceedings{shcherba2025mops,
  title     = {Meta-Optimization and Program Search using Language Models for Task and Motion Planning}, 
  author    = {Denis Shcherba and Eckart Cobo-Briesewitz and Cornelius V. Braun and Marc Toussaint},
  year      = {2025},
  booktitle = {Conference on Robot Learning}
}