DeepRobots

Deep Reinforcement Learning for Robot Locomotion

Contributors

Jack Shi (junyaoshi)
Tony Dear (tonydear)

Dependencies

Python ≥ 3.5
Keras 2.2.4
matplotlib 3.1.0

If python requirement is not met, a virtual environment can be installed using the follwoing method on terminal:

conda create -n py37 python=3.7 anaconda 
conda activate py37
pip install keras=2.2.4
pip install matplotlib=3.1.0

Directories

DiscreteRL: different versions of implementation of discrete RL agent
DQN: most up-to-date implementation of DQN agent
- DQN_agent.py: a general-purpose class that can be easily used to train different kinds of snake robots.
- DQN_runner.py: a simple interface for running a single trial of DQN with customized parameters
- DQN_multi_runner.py: a simple interface for running a multiple trials of DQN with customized parameters
DQN_old: some old implementation of DQN, please ignore
Mathematica Notebooks: used for generating animation and plots
Robots: different types of snake robot models.
- DiscreteDeepRobots.py, ContinuousDeepRobot.py: wheeled snake robot implementation
- ContinuousSwimmingBot.py: swimming snake robot implementation

Quick Start Guide

Clone this repository into some local directory local_dir on the local machine.

First, open terminal and change directory:

cd local_dir

Clone this repository in the local directory:

git clone https://github.com/junyaoshi/DeepRobots.git

Then, change directory to DeepRobots directory:

cd DeepRobots

Make sure the system path is set up properly so that all library and module imports can be successfully performed. See DQN/DQN_runner.py for example.

To perform a single DQN trial, run:

python DQN/DQN_runner.py

To perform a multiple different trials of DQN, run:

python DQN/DQN_multi_runner.py

Physical Experiments

To perform DQN physical experiments with the physical robot, first change directory to DeepRobots directory:

cd DeepRobots

Then run:

python DQN/DQN_physical_runner.py

Jesse-Redford / DeepRobots