Maltimore / robotic_nanofabrication

Code for the paper "Autonomous robotic nanofabrication with reinforcement learning"

Robotic nanofabrication

Introduction

This repository contains the code for the Reinforcement Learning agent described in the paper

Autonomous robotic nanofabrication with reinforcement learning

Science Advances (open access): https://advances.sciencemag.org/content/6/36/eabb6987.full

Arxiv: https://arxiv.org/abs/2002.11952

Usage

In order to install the requirements (using conda):

conda env create -f conda_environment.yaml
conda activate robotic_nanofab

In order to run the RL algorithm in the simulated environment, do

python robotic_nanofabrication/main.py --output_path=outfiles/test1

Recreate swarm plots in paper

The swarm plots in the paper (Fig. 2D) have been created using the follwing repository:

https://github.com/maltimore/gridsearch_helper

you should use tag 0.0.2.

That repository has the scripts to parallelize experiments on the Sun Grid Engine or Univa Grid Engine. Additionally, it contains python files that can aggregate the data from the parallelized runs and create the swarm plots.