No project description provided
Rex: an open-source domestic robot
The goal of this project is to train an open-source 3D printed quadruped robot exploring
Reinforcement Learning and
OpenAI Gym. The aim is to let the robot learns domestic and generic tasks in the simulations and then
successfully transfer the knowledge (
Control Policies) on the real robot without any other manual tuning.
This project is mostly inspired by the incredible works done by Boston Dynamics.
rexctl - A CLI application to bootstrap and control Rex running the trained
rex-cloud - A CLI application to train Rex on the cloud.
Rex-gym: OpenAI Gym environments and tools
This repository contains different
OpenAI Gym Environments used to train Rex, the Rex URDF model,
the learning agent and some scripts to start the training session and visualise the learned
The CLI application allows batch training, policy reproduction and rendered single training sessions.
Python 3.7 virtual environment, e.g. using
conda create -n rex python=3.7 anaconda conda activate rex
Install the public
pip install rex_gym
Install from source
Alternately, clone this repository and run from the root of the project:
pip install .
rex-gym --help to display the available commands and
rex-gym COMMAND_NAME --help to show the help
message for a specific command.
Policy player: run a pre-trained agent
To start a pre-trained agent (play a learned
rex-gym policy --env ENV_NAME
|Turn (on spot)||turn||
|init_orient||The starting orientation in rad.|
|target_orient||The target orientation in rad.|
|log-dir||The path where the log directory will be created. (Required)|
|playground||A boolean to start a rendered single training session|
|agents-number||Set the number of parallel agents|
Run a single training simulation
To start a rendered single training session (
rex-gym train --playground True --env ENV_NAME --log-dir LOG_DIR_PATH
Start a new batch training simulation
To start a new batch training session:
rex-gym train --env ENV_NAME --log-dir LOG_DIR_PATH
PPO Agent configuration
You may want to edit the PPO agent's default configuration, especially the number of parallel agents launched during the simulation.
--agents-number flag, e.g.
This configuration will launch 10 agents (threads) in parallel to train your model.
The default value is setup in the
def default(): """Default configuration for PPO.""" # General ... num_agents = 20
I've printed the components using a Creality Ender3 3D printer, with PLA and TPU+.
The idea is to extend the robot adding components like a robotic arm on the top of the rack and a LiDAR sensor.
Rex is a 12 joints robot with 3 motors (
Foot) for each leg.
poses signals (see
/model/rex.py) set the 12 motor angles and allow Rex to stand up.
The robot model is imported in
pyBullet using an URDF file.
This is the list of tasks this experiment will cover:
- Basic controls
- Gallop/Walk straight on - forward/backward
- Turn left/right on the spot
- Stand up/Sit down
- Side swipe
- Fall recovery
- Reach a specific point in a map
- Grab an object
Basic Controls: Run
Goal: how to run straight on.
There is a good number of papers on quadrupeds locomotion, some of them with sample code. Probably, the most complete collection
of examples is the Minitaur folder in the Bullet3 repository.
For this task, the
Minitaur Reactive Environment explained in the paper Sim-to-Real: Learning Agile Locomotion For Quadruped Robots
is a great example.
Galloping gait - from scratch
In this very first experiment, I let the system learn from scratch: giving the feedback component large output bounds
leg model (see
galloping_env.py) forces legs and foots movements (positive or negative direction, depending on the leg) influencing the learning
score and time. In this first version, the
leg model holds the Shoulder motors in the start position (0 degrees).
As in the Minitaur example, I'm using the Proximal Policy Optimization (PPO).
The emerged galloping gait shows the chassis tilled up and some unusual positions/movements (especially starting from the initial pose) during the locomotion. The
leg model needs improvements.
Galloping gait - bounded feedback
To improve the gait, in this second simulation, I've worked on the
I set bounds for both
Foot angles, keeping the
Shoulder in the initial position.
The emerged gait now looks more clear.
Galloping gait - balanced feedback
Another test was made using a balanced feedback:
The Action Space dimension is equals to 4, the same angle is assigned to both the front legs and a different one to the rear ones. The very same was done for the foot angles.
The simulation score is massively improved (about 10x) as the learning time while the emerged gait is very similar to the
bounded feedback model.
The Tensorflow score with this model, after ~500k attempts, is the same after ~4M attempts using any other models.
Basic Controls: Walk
Goal: how to walk straight on.
Starting from Minitaur Alternating Leg
environment, I've used a sinusoidal signal as
leg_model alternating the Rex legs during the locomotion. The feedback component has small
bounds [-0.1,0.1] as in the original script.
Basic Controls: Turn left/right
Goal: How to reach a certain orientation turning on the spot.
In this environment the
leg_model applies a 'steer-on-the-spot' gait, allowing Rex to moving towards a specific orientation.
The reward function takes the chassis position/orientation and compares it with a fixed target position/orientation.
When this difference is less than 0.1 radiant, the
leg_model is set to the stand up. In order to make the learning more robust,
the Rex starting orientation is randomly chosen (every 'Reset' step).
Basic Controls: Stand up
Goal: Reach the base standing position starting from the rest position
This environment introduces the
rest_postion, ideally the position assumed when Rex is in stand-by.
leg_model is the
stand_low position, while the
signal function applies a 'brake' forcing Rex to assume an halfway position
before completing the movement.
Sim-to-Real: Learning Agile Locomotion For Quadruped Robots and all the related papers. Google Brain, Google X, Google DeepMind - Minitaur Ghost Robotics.
Deok-yeon Kim creator of SpotMini.
The great work in rendering the robot platform done by the SpotMicroAI community.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size rex_gym-0.1.9.tar.gz (23.5 MB)||File type Source||Python version None||Upload date||Hashes View|