Kimera-VIO Offline Installation Guide

Platform: Ubuntu 20.04.6 LTS (amd64)
Target: Compile and run Kimera-VIO on an offline (air-gapped) Ubuntu machine

Overview

This guide covers the full process of installing all dependencies and compiling Kimera-VIO on an offline Ubuntu 20.04 machine. All packages and third-party libraries are pre-built on an online machine, transferred, and installed locally.

Prerequisites

Ensure the following files have been prepared on an online Ubuntu 20.04 machine and transferred to the offline machine:

Step 1: Unpack Cached APT Packages

please copy all files of var-lib-apt-lists folder into /var/lib/apt/lists
please copy all deb files of var-cache-apt-archives folder into /var/cache/apt/archives

## Step 2: Install System Dependencies

```bash
sudo apt update --no-download
sudo apt-get install -y --no-install-recommends apt-utils
sudo apt-get install -y cmake
sudo apt-get install -y libboost-all-dev
sudo apt-get install -y build-essential unzip pkg-config \
  libjpeg-dev libpng-dev libtiff-dev libvtk7-dev libgtk-3-dev \
  libparmetis-dev libatlas-base-dev gfortran
sudo apt-get install -y libtbb-dev
sudo apt-get install -y libgflags-dev libgoogle-glog-dev

Step 3: Install Third-Party Libraries

3.1 GTSAM (>= 4.1)

tar -xvf gtsam-install.tar
sudo cp -r gtsam-install/* /usr/local/
sudo ldconfig

3.2 OpenCV 4.5.1

tar -xvf opencv-4.5.1-install.tar
sudo cp -r opencv-4.5.1-install/* /usr/local/
sudo ldconfig

3.3 OpenGV

tar -xvf opengv-install.tar
sudo cp -r opengv-install/* /usr/local/
sudo ldconfig

3.4 DBoW2

tar -xvf DBoW2-install.tar
sudo cp -r DBoW2-install/* /usr/local/
sudo ldconfig

3.5 Kimera-RPGO

tar -xvf Kimera-RPGO-install.tar
sudo cp -r Kimera-RPGO-install/* /usr/local/
sudo ldconfig

Step 4: Compile Kimera-VIO

cd Kimera-VIO
mkdir build
cd build
cmake -DKIMERA_BUILD_TESTS=OFF \
  -DDBoW2_INCLUDE_DIR=/usr/local/include \
  -DDBoW2_LIBRARY=/usr/local/lib/libDBoW2.so ..
make -j$(nproc)

Step 5: Run Kimera-VIO

bash ./scripts/stereoVIOEuroc.bash

Before running, edit the script to set the dataset path:

# The path can be absolute, or relative to the script location.
DATASET_PATH="/home/ubuntu/Desktop/Work/dataset/MH_01_easy"