Skip to content

Latest commit

 

History

History
134 lines (93 loc) · 5.19 KB

getting_started.rst

File metadata and controls

134 lines (93 loc) · 5.19 KB
Raw

Getting Started

Prerequisites

  1. Operating System:
    Windows 10 and Mac(!) both works!
  2. Rhinoceros 3D 6.0
    We will use Rhino / Grasshopper as a frontend for inputting geometric and numeric paramters, and use various python packages as the computational backends. The tight integration between Grasshopper and python environments is made possible by COMPAS and COMPAS_FAB.
  3. Git
    We need Git for fetching required packages from github.
  4. Miniconda
    We will install all the required python packages using Miniconda (a light version of Anaconda). Miniconda uses environments to create isolated spaces for projects' depedencies.
  5. Microsoft Visual Studio Build tools
    Our simulation environment pybullet has a C++ backend, which needs Microsoft Visual C++ 14.0 to compile and build the python bindings. Note that this is needed only for Windows OS.

Working in a conda environment

It is recommended to set up a conda environment to create a clean, isolated space for installing all the required python packages.

Type in the following commands in your Anaconda terminal to create a conda environment (search for Anaconda Prompt in the Windows search bar):

git clone --recursive https://github.com/yijiangh/coop_assembly.git
cd coop_assembly
conda create -n cp_ws python=3.7
conda activate cp_ws

Notice that we have cloned this repository using the --recursive flag (if not then type git submodule update --init --recursive) to make sure we have all the git submodules in place.

The last two lines above will create a conda environment called cp_ws. Now, we install all the depedency packages:

pip install ./external/pybullet_planning
pip install -e .

This will install the main depedency pybullet_planning. The last line install coop_assembly in a debug mode, which means your changes will be directly reflected in your execution (from python or from GHPython invokes).

If you see an error message like Error: Microsoft Visual C++ 14.0 is required, please see troubleshooting for instructions to install the Microsoft Visual Studio Build tools.

Great - we are almost there! Now type python in your Anaconda Prompt, and test if the installation went well:

>>> import compas
>>> import compas_fab
>>> import pybullet
>>> import coop_assembly

If that doesn't fail, you're good to go! Exit the python interpreter (either typing exit() or pressing CTRL+Z followed by Enter).

Now let's make all the installed packages available inside Rhino. Still from the Anaconda Prompt, type the following:

In order to make coop_assembly accessible in Rhino/Grasshopper, we need the run the following commands in the Anaconda prompt first and then restart Rhino:

python -m compas_rhino.install
python -m compas_rhino.install -p coop_assembly

And you should be able to see outputs like:

Installing COMPAS packages to Rhino 6.0 IronPython lib:
IronPython location: C:\Users\<User Name>\AppData\Roaming\McNeel\Rhinoceros\6.0\Plug-ins\IronPython (814d908a-e25c-493d-97e9-ee3861957f49)\settings\lib

compas               OK
compas_rhino         OK
compas_ghpython      OK
coop_assembly        OK

Completed.

Congrats! 🎉 You are all set!

Grasshopper examples can be found in the examples folder. For Stefana and her students, please see examples/shape_gen_GH for the latest examples on design generation.

Installating sequence and motion planning stack

We have to install a few additional packages to make our sequence and motion planning stack to work.

For the regression searcha algorithm, you will need to install conmech for structural analysis and ikfast_pybind for analytic IK.

pip install ./external/conmech
pip install ./external/ikfast_pybind

Installating PDDLStream (WIP)

To install PDDLStream and pyplanner for using semantic attachments in the incremental algorithm, we have to add the path to externals/pyplanner to the environment variable PYPLANNERS_PATH.

Then, cd externals/pddlstream/FastDownward and follow the instructions here to install FastDownward, the default discrete AI planner used by PDDLStream.