rlober / task-optim

Optimizing Task Feasibility using Model-Free Policy Search and Model-Based Whole-Body Control

task-optim

Under serious construction...

Please bear with me while I clean this up.

Dependencies & Install

pip/pip3 install DIRECT emcee numpy scipy matplotlib cma sklearn

RoBO fork:

git clone -b corl_2017 https://github.com/rlober/RoBO
cd RoBO
pip/pip3 install .

Code:

git clone https://github.com/rlober/ra-l_2017.git
cd ra-l_2017
mkdir build
cd build
cmake ..
make install

git clone https://github.com/rlober/gttraj.git
cd gttraj
mkdir build
cd build
cmake ..
make install

git clone https://github.com/rlober/com-traj-gen.git
cd com-traj-gen
mkdir build
cd build
cmake ..
make install

Gazebo:

git clone -b corl_2017 https://github.com/robotology-playground/icub-gazebo-wholebody.git

WBIController:

cd $CODYCO_SUPERBUILD_ROOT
cd main/WBIToolboxControllers
git checkout corl_2017
cd ../../build
make

env variables:

export PYTHONPATH=$PYTHONPATH:[path]/ra-l_2017/optimization-scripts
export GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:[path]/icub-gazebo-wholebody

Generate python documentation:

cd [path_to_repo]/optimization-scripts/docs
sphinx-apidoc -f -o _source/ ../task_optim/
make html