uoa-cares/nasa_td3

Official PyTorch Implementation of Our Paper Image-Based Deep Reinforcement Learning with Intrinsically Motivated Stimuli: On the Execution of Complex Robotic Tasks

NaSA_TD3

Official PyTorch Implementation of Image-Based Deep Reinforcement Learning with Intrinsically Motivated Stimuli: On the Execution of Complex Robotic Tasks

General Overview

See our Paper-Blog for details for pseudocode with more details of the training process as well as details of hyperparameters, full source code and videos of each task.

Prerequisites

Library	Version
Our RL Support Libray	link
DeepMind Control Suite	link

Network Architecture

Instructions Training

To train the NaSA-TD3 algorithm on the deep mind control suite from image-based observations, please run:

python3 train_loop.py --env=ball_in_cup --task=catch --seed=1 --intrinsic=True

Our Results

Citation

If you use either the paper or code in your paper or project, please kindly star this repo and cite our work.