Home

Welcome to the RLG Wiki!

Here you can find more detailed information about the Robot Learning Gym.

Installation

Clone this repository and run sudo pip install . in the home directory.

About

RLG deals with three facets of robot learning:

1. Tasks
2. Robots
3. Agents

The idea is that RLG should provide users with different tasks and robots and a few basic agents, enabling users to start creating more intelligent agents.

Tasks/Robots/Agents

Tasks:

1. Door (custom made)
2. Beam balance
3. Lift Can (using MuJoCo example XML file)
4. Obstacle Course
5. Snake

Robots (many of these were taken from the MuJoCo Resources, although some needed slight edits to work):

1. Atlas (Virtual and v5)
2. Barrett hand
3. Baxter
4. Claw/hand
5. Darwin
6. Jaco
7. PR2
8. Sandia Hand

Agents: DefaultAgent simply samples from the action space (i.e: acts randomly)

---

Tested on Ubuntu 14.04 LTS