For this tutorial session, we are going to need:
- Crocoddyl v2 (2.0.2)
- Pinocchio
- For some chapters: Casadi with the Pinocchio bindings. Pinocchio v2.99 is then recommanded.
We are going to use the Conda channel prepared for the Agimus Winter School. All the required packages are available on the AWS channel. Conda is to be installed on your machine by following these instructions.
You can then install a package by typing in your terminal:
conda create -n wan24 python=3.10
conda activate wan24
conda install -c agm-ws-2023 -c conda-forge mim-solvers
The two first lines create a new environment named wan24
and then activate it.
The third line installs mim-solvers
using the AWS channel where the packages have been compiled (MiM-solvers is the most down-stream package, which then drains Pinocchio, Crocoddyl, etc).
You then need to install the following additional tools via pip:
pip install tqdm meshcat ipython jupyterlab ipywidgets matplotlib gepetuto`
You can install with PyPI but you then need two environments:
- the first one with the Pinocchio Casadi module, specifically compiled and frozen to a 2023 version
- the second one with the mainstream Crocoddyl and Pinocchio.
This can be done using two virtual environments.
-
create an environment:
python -m venv .venv-pincasadi
-
activate it:
source .venv-pincasadi/bin/activate
-
update pip:
pip install -U pip
- install dependencies:
pip install example-robot-data-jnrh2023 jupyterlab meshcat ipywidgets matplotlib gepetuto
-
when done with the virtual environment, deactivate:
deactivate
-
create an environment:
python -m venv .venv-croc
-
activate it:
source .venv-croc/bin/activate
-
update pip:
pip install -U pip
-
install dependencies:
pip install cmeel-mim-solvers jupyterlab meshcat ipywidgets matplotlib gepetuto
-
when done with the virtual environment, deactivate:
deactivate
The tutorials are all in Jupyter Lab. Start them with PYTHONPATH=. jupyter-lab