Installing Kalibr via Docker or from source

Kalibr can be installed in two ways: using Docker (recommended for most users) or building from source inside a ROS catkin workspace. Docker is simpler because it bundles all dependencies — including the correct ROS distribution, Python version, and native libraries — into a reproducible image.

System requirements

OS: Ubuntu 16.04, 18.04, or 20.04 (64-bit)
ROS: ROS 1 (Kinetic, Melodic, or Noetic depending on Ubuntu version)
Docker: Required for the Docker installation path
RAM: 8 GB or more recommended for calibration with large datasets

Docker (recommended)
Source build

Docker is the easiest way to get a fully working Kalibr environment. The provided Dockerfiles build a self-contained image with all system dependencies, ROS, and the compiled Kalibr workspace.

Install Docker

If Docker is not already installed, follow the official Docker installation guide for your Ubuntu version. After installing, verify it works:

docker run hello-world

Clone the Kalibr repository

git clone https://github.com/ethz-asl/kalibr.git
cd kalibr

Build the Docker image

Choose the Dockerfile that matches your target Ubuntu version. Ubuntu 20.04 is recommended.

Ubuntu 20.04 (recommended)
Ubuntu 18.04
Ubuntu 16.04

docker build -t kalibr -f Dockerfile_ros1_20_04 .

This builds from osrf/ros:noetic-desktop-full and installs:

Python 3 with scipy, matplotlib, and related packages
Eigen3, Boost, SuiteSparse, OpenCV, TBB, BLAS, LAPACK
python3-catkin-tools for the build system
Kalibr compiled in Release mode inside /catkin_ws

docker build -t kalibr -f Dockerfile_ros1_18_04 .

Builds from osrf/ros:melodic-desktop-full with Python 2/3 dependencies.

docker build -t kalibr -f Dockerfile_ros1_16_04 .

Builds from osrf/ros:kinetic-desktop-full with Python 2 dependencies.

The build compiles all C++ libraries with -j$(nproc) threads. On a 4-core machine, expect 15–30 minutes for the first build.

Run the container

Mount a local directory into the container so you can pass ROS bags in and retrieve calibration results without copying files manually.

docker run -it \
  -e "DISPLAY" \
  -e "QT_X11_NO_MITSHM=1" \
  -v /tmp/.X11-unix:/tmp/.X11-unix:rw \
  -v "$PWD/data:/data" \
  kalibr

The container’s ENTRYPOINT sets KALIBR_MANUAL_FOCAL_LENGTH_INIT=1 and drops you into a shell inside /catkin_ws with the ROS environment ready to use.

Place your ROS bag files and target YAML files in the data/ directory on your host machine. They will be available at /data inside the container.

Source the workspace and verify

Inside the running container, source the built workspace and confirm the tools are available:

source /catkin_ws/devel/setup.bash
kalibr_calibrate_cameras --help

You should see the help output for the camera calibration tool.

Building from source gives you the most control and is useful when you need to modify Kalibr’s code or integrate it into an existing ROS workspace.

Install ROS

Install a supported ROS 1 distribution for your Ubuntu version:

Ubuntu	ROS distribution
20.04	ROS Noetic
18.04	ROS Melodic
16.04	ROS Kinetic

Follow the official ROS installation instructions for your distribution.

Install system dependencies

Install the required libraries. This example uses ROS Noetic on Ubuntu 20.04:

sudo apt-get update && sudo apt-get install -y \
  git wget autoconf automake \
  python3-dev python3-pip python3-scipy python3-matplotlib \
  ipython3 python3-wxgtk4.0 python3-tk python3-igraph python3-pyx \
  libeigen3-dev libboost-all-dev libsuitesparse-dev \
  doxygen \
  libopencv-dev \
  libpoco-dev libtbb-dev libblas-dev liblapack-dev libv4l-dev \
  python3-catkin-tools python3-osrf-pycommon

For Ubuntu 18.04 (Melodic), replace python3-* packages with the appropriate python-* equivalents, and replace python3-catkin-tools with python-catkin-tools. For Ubuntu 16.04 (Kinetic), use python2.7-dev and python-wxgtk3.0.

Create a catkin workspace

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws
catkin init
catkin config --extend /opt/ros/noetic
catkin config --cmake-args -DCMAKE_BUILD_TYPE=Release

Adjust /opt/ros/noetic to match your installed ROS distribution (melodic or kinetic).

Clone Kalibr into the workspace

cd ~/catkin_ws/src
git clone https://github.com/ethz-asl/kalibr.git

Build Kalibr

cd ~/catkin_ws
catkin build -j$(nproc)

The build compiles several C++ libraries. Allow 15–30 minutes on a typical workstation. Use -j4 or a lower number if you run out of memory.

Source the workspace

Add the workspace to your shell environment. Run this once to add it permanently:

echo "source ~/catkin_ws/devel/setup.bash" >> ~/.bashrc
source ~/.bashrc

Or source it manually for the current session:

source ~/catkin_ws/devel/setup.bash

Verify the installation

kalibr_calibrate_cameras --help

A successful installation prints the usage information for the camera calibration tool.

Set export KALIBR_MANUAL_FOCAL_LENGTH_INIT=1 in your shell to enable the manual focal length fallback if automatic initialization fails.

Troubleshooting

Docker build fails on apt-get install

Run docker build --no-cache to force a fresh package list fetch, or check that your host has network access during the build.

catkin build fails with 'Could not find package'

Verify that catkin config --extend points to the correct ROS installation path and that you sourced the ROS setup script (source /opt/ros/<distro>/setup.bash) before running catkin build.

kalibr_calibrate_cameras: command not found

You need to source the workspace before using Kalibr commands. Run source ~/catkin_ws/devel/setup.bash (source build) or source /catkin_ws/devel/setup.bash (Docker).

Focal length initialization fails

Set export KALIBR_MANUAL_FOCAL_LENGTH_INIT=1 before running any calibration command. The Docker image sets this automatically.

Get Started

Calibration Guides

Calibration Targets

Concepts

Installing Kalibr via Docker or from source

System requirements

Troubleshooting

Next steps

Quickstart

Calibration targets

Build docs developers (and LLMs) love

Get Started

Calibration Guides

Calibration Targets

Concepts

Documentation Index

​System requirements

​Troubleshooting

​Next steps

Quickstart

Calibration targets

Build docs developers (and LLMs) love

System requirements

Troubleshooting

Next steps