Skip to content

README.md

Latest commit

 

History

History
105 lines (71 loc) · 2.63 KB

README.md

File metadata and controls

105 lines (71 loc) · 2.63 KB

LineFlow

PyPI

LineFLow is a python framework to simulate assembly lines. It allows to model arbitrary discrete part assembly lines and provides an gymnasium environment to optimize them with reinforcement learning. The documentation can be found here.

til

Install

Install with

pip install lineflow-rl

Examples

Visualization

This is how an assembly line can be implemented and visualized:

from lineflow.simulation import Line, Source, Sink, Process

class SimpleLine(Line):

    def build(self):

        # Set up stationary objects
        source = Source(
            name='Source',
            processing_time=5,
            position=(100, 500),
            unlimited_carriers=True,
        )

        process = Process('Process', processing_time=6, position=(350, 500))
        sink = Sink('Sink', processing_time=3, position=(600, 500))
        
        # Wire them with buffers
        source.connect_to_output(station=process, capacity=3)
        process.connect_to_output(station=sink, capacity=2)


line = SimpleLine()
line.run(simulation_end=500, visualize=True)

df = line.get_observations()

Training RL agents

This is how an RL agent can be trained using LineFlow:

from stable_baselines3 import PPO
from lineflow.simulation import LineSimulation

line = SimpleLine()
env = LineSimulation(line, simulation_end=100)
model = PPO("MlpPolicy", env)
model.learn(total_timesteps=100)

Docs

Serve the docs with

mkdocs serve

Paper

If you use our work in your research, please consider citing us with

@InProceedings{pmlr-v267-muller25c,
  title = {{L}ine{F}low: A Framework to Learn Active Control of Production Lines},
  author = {M\"{u}ller, Kai and Wenzel, Martin and Windisch, Tobias},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  pages = {45212--45235},
  year = {2025},
  editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry},
  volume = {267},
  series = {Proceedings of Machine Learning Research},
  month = {13--19 Jul},
  publisher = {PMLR},
  url = {https://proceedings.mlr.press/v267/muller25c.html},
}

See this README for more details how to run the benchmarks.

Funding

The research behind LineFlow is funded by the Bavarian state ministry of research. Learn more here.