REx Hopper

Table of Contents

• Intro
• Setup
• Info

---

Intro

This repository mainly contains Python code for simulation of the REx Hopper in PyBullet. There are also supplementary Jupyter notebooks and matlab files in /scripts.

---

Setup

1. Clone this directory wherever you want.

2. Make sure both Python 3.8 and pip are installed.

sudo apt install python3.8
sudo apt-get install python3-pip
python3.8 -m pip install --upgrade pip

2. I recommend setting up a virtual environment for this, as it requires the use of a number of specific Python packages.

sudo apt-get install python3.8-venv
cd rex_hopper_python/src
python3.8 -m venv env

For more information on virtual environments: https://docs.python.org/3/library/venv.html

3. Activate the virtual environment, and then install numpy, scipy, matplotlib, sympy, transforms3d, pybullet, cvxpy, argparse, and more.

source env/bin/activate
python3.8 -m pip install numpy scipy matplotlib sympy transforms3d pybullet cvxpy argparse dill SciencePlots

Don't use sudo here if you can help it, because it may modify your path and install the packages outside of the venv.

Here is the PyBullet tutorial.

---

Info

Here is some example code, which runs a simulation of a hopper standing with reaction wheels:

cd rex_hopper_python/src
source env/bin/activate
python3.8 run.py design_rw ik_static --N_run=5000 --plot

Argparse Arguments

There are two required argparse arguments: model and control. In addition, there are a number of optional arguments.

model

Choices:

• design_rw (The design-based model with reaction wheels, unconstrained with the world frame.)
• design_rw_alt

ctrl

Choices:

• mpc (model predictive controller)
• wbc_raibert (raibert hopping with whole body leg controller)
• wbc_vert (hopping straight up and down with whole body leg controller)
• wbc_static (standing still with whole body leg controller)
• ik_vert (Jumps up and down using simple inverse kinematics and PID.)
• ik_static (Stands using simple inverse kinematics and PID.)

optional arguments

• --plot (Use Matplotlib to plot parameters.)
• --fixed (Fixes the base of the robot to the world frame.)
• --spring (Adds a parallel spring to a parallel leg. Does not work with serial or belt models.)
• --record (Records video of sim.)
• --recalc (Recalculate leg kinematics, dynamics, jacobian, etc. Saves new solution for future runs.)
• --direct (Turn off gravity.)