data_collection_pipeline

This is a pipeline to collect data from human demonstrations for the Franka robot using an Intel RealSense D415 camera.

Installation

1. Install mamba (if you don't already have it).

2. Create the conda environment named collector_env:

mamba env create -f environment.yml

3. Activate the environment:

mamba activate collector_env

4. Install libfranka v0.9.2 (required for this project):

sudo apt-get update
sudo apt-get install -y build-essential cmake git libpoco-dev libeigen3-dev libfmt-dev
git clone --branch 0.9.2 --recursive https://github.com/frankaemika/libfranka libfranka-0.9.2
cd libfranka-0.9.2/
mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Release \
	-DBUILD_TESTS=OFF \
	-DCMAKE_INSTALL_PREFIX=$HOME/.local \
	-DBUILD_EXAMPLES=ON
make -j$(($(nproc)-1))

5. Build the data_collection project:

cd ../..
cd data_collection
mkdir build && cd build
cmake -S . -B build \
	-DCMAKE_BUILD_TYPE=Release \
	-DCMAKE_PREFIX_PATH=$HOME/.local \
	-DFranka_DIR=$HOME/.local/lib/cmake/Franka
cmake --build build -j