Skip to content

Latest commit

 

History

History
99 lines (73 loc) · 3.14 KB

INSTALL.md

File metadata and controls

99 lines (73 loc) · 3.14 KB
Raw

Installation on Ubuntu

Prerequisites

  • Ubuntu 16.04/18.04/20.04 with compatible ROS version
  • Python >= 3.8
  • gcc & g++ >= 5.4
  • OpenCV 3 or 4
  • Nvidia GPU (with compatible cuda toolkit and cuDNN) if want to run online segmentation
  • Anaconda for configuring python dependencies

Clone the repository & install catkin dependencies

First create and navigate to your catkin workspace

cd <your-working-directory>
mkdir <your-ros-ws>/src && cd <your-ros-ws>

Then, initialize the workspace and configure it. (Remember to replace by your ros version)

catkin init
catkin config --extend /opt/ros/<your-ros-version> --merge-devel 
catkin config --cmake-args -DCMAKE_CXX_STANDARD=14 -DCMAKE_BUILD_TYPE=Release

Download this repository to your ROS workspace src/ folder with submodules via:

cd src
git clone --recursive https://github.com/hmz-15/Interactive-Scene-Reconstruction.git

Then add dependencies specified by .rosinstall using wstool. For your conivenience, we provide a script to automatically setup the wstool configuration. Please make sure you run the script at the root of your ros workspace.

cd <your-ros-ws>
sh src/Interactive-Scene-Reconstruction/wstool_setup_https.sh

Noted that switch to wstool_setup_ssh.sh if you are using ssh instead of https.

Install python dependencies

We assume using conda virtual environment to configure python dependencies. This requires Anaconda to be installed and initialized as prerequisite. We create a conda env with python3.7 and install the dependencies:

conda create --name robot-scene-recon python=3.7 -y
conda activate robot-scene-recon
pip install pip --upgrade

# detectron2 dependencies
pip install torch torchvision
python -m pip install detectron2 -f \
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu102/torch1.10/index.html
# other dependencies
cd src/Interactive-Scene-Reconstruction
pip install -r requirements.txt

Note that you may need to adjust the version of torch, torchvision and detectron2 based on your cuda/cuDNN version.

You can deactivate the conda env using:

conda deactivate

Build packages

We first build the python package containing panoptic segmentation server.

conda activate robot-scene-recon
cd mapping/rp_server
make dev

Then we build the ros packages with catkin build.

cd <your-ros-ws>
catkin build panoptic_mapping_pipeline map_proc scene_builder gazebo_simulation -j2
source devel/setup.bash

Alternatively, you may want to build packages for each individule module:

# Build packages for panoptic mapping
catkin build panoptic_mapping_pipeline -j2
# Build packages for cad replacement 
catkin build map_proc
# Build packages for interactive scene reconstruction & gazebo demo
catkin build scene_builder gazebo_simulation

Please replace bash by zsh if zsh is your default shell.