Skip to content

Python code for Deep Reinforcement Learning with Transfer Learning in a Simulated Environment

License

Notifications You must be signed in to change notification settings

aqeelanwar/DRLwithTL

Repository files navigation

NOTE: This repository will no longer be updated and has been made a part of a larger repository PEDRA. You are advised to used PEDRA instead of this repository.

Deep Reinforcement Learning with Transfer Learning - Simulated Drone and Environment (DRLwithTL-Sim)

What is DRLwithTL-Sim?

This repository uses Transfer Learning (TL) based approach to reduce on-board computation required to train a deep neural network for autonomous navigation via Deep Reinforcement Learning for a target algorithmic performance. A library of 3D realistic meta-environments is manually designed using Unreal Gaming Engine and the network is trained end-to- end. These trained meta-weights are then used as initializers to the network in a simulated test environment and fine-tuned for the last few fully connected layers. Variation in drone dynamics and environmental characteristics is carried out to show robustness of the approach. The repository containing the code for real environment on a real DJI Tello drone can be found @ DRLwithTL-Real

Cover Photo Cover Photo Cover Photo

Introductory Video

Watch the video

Installing DRLwithTL

For detailed instructions on how to install, configure and run DRLwithTL, please refer PEDRA

Citing

If you find this repository useful for your research please use the following bibtex citations

@ARTICLE{2019arXiv191005547A,
       author = {Anwar, Aqeel and Raychowdhury, Arijit},
        title = "{Autonomous Navigation via Deep Reinforcement Learning for Resource Constraint Edge Nodes using Transfer Learning}",
      journal = {arXiv e-prints},
     keywords = {Computer Science - Machine Learning, Statistics - Machine Learning},
         year = "2019",
        month = "Oct",
          eid = {arXiv:1910.05547},
        pages = {arXiv:1910.05547},
archivePrefix = {arXiv},
       eprint = {1910.05547},
 primaryClass = {cs.LG},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2019arXiv191005547A},
      adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}
@article{yoon2019hierarchical,
  title={Hierarchical Memory System With STT-MRAM and SRAM to Support Transfer and Real-Time Reinforcement Learning in Autonomous Drones},
  author={Yoon, Insik and Anwar, Malik Aqeel and Joshi, Rajiv V and Rakshit, Titash and Raychowdhury, Arijit},
  journal={IEEE Journal on Emerging and Selected Topics in Circuits and Systems},
  volume={9},
  number={3},
  pages={485--497},
  year={2019},
  publisher={IEEE}
}

Authors

License

This project is licensed under the MIT License - see the LICENSE.md file for details

About

Python code for Deep Reinforcement Learning with Transfer Learning in a Simulated Environment

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages