arm architecture, Xavier NX DK, Ubuntu 18 04. ROS melocdic drives Azure Kinect dk camera

arm architecture, Xavier NX DK, Ubuntu 18 04. ROS melocdic drives Azure Kinect dk camera

Device Xavier NX DK azure Kinect DK camera
System: Ubuntu 18 04 ros melodic

Devices under X86 architecture drive Azure cameras. You can read a blog:
x86 architecture, Ubuntu 18 04. ROS melocdic drives Azure Kinect dk camera
Preface: Xavier NX DK is an arm product of nvidia and has its own CUDA acceleration function. Therefore, this time, the Azure Kinect dk camera is driven under the arm architecture. In terms of SDK configuration, it is slightly different from x86.
Reference blog:
nvidia agx xavier runs azure kinect DK ros

ubuntu18.04 driver official SDK

1. Download the official SDK package

Add code warehouse:

curl -sSL | sudo apt-key add -

Update software source under arm:

sudo apt-get update -y && apt-get install curl -y

Add warehouse under arm64

sudo apt-add-repository 
sudo apt-get update

Add a warehouse under x86. There is no need to add a warehouse when using arm series embedded board

sudo apt-add-repository

Source code installation: because the subsequent projects use sdk1 Version 4, so version 1.4 is installed here

git clone -b v1.4.0 download

Because the arm architecture is used, is installed here so. 2.0 is best obtained from the files downloaded from the arm source

sudo apt install k4a-tools
sudo apt install libk4a1.4-dev      #Note that there are many versions of libk4a. The 1.4 version selected here can be installed according to the selection of SDK

Copy to / usr / lib / x86_ 64 Linux GNU directory

cd /usr/lib/aarch64-linux-gnu/libk4a1.4
sudo cp /usr/lib/aarch64-linux-gnu

If it is under x86 architecture, copy it to / usr / lib / x86_ 64 Linux GNU directory

cd  /usr/lib/x86_64-linux-gnu
sudo cp /usr/lib/x86_64-linux-gnu

3. Compile and install azure Kinect sensor SDK

cd   Azure-Kinect-Sensor-SDK
mkdir build
cd build
cmake .. -GNinja -DCMAKE_BUILD_TYPE=Debug

ps:cmake... - GNinja is the configuration for generating ninja. Ninja is a small compilation system with parallel compilation and high speed

When finished, execute

sudo ninja install

Problems encountered when executing cmake... - GNinja:


Solution: add source

sudo apt-add-repository 
sudo apt-get update


sudo apt install libsoundio-dev
cmake .. -GNinja


sudo apt-get install  libudev-dev 

4. Start k4aviewer

Connect the camera, still in the build folder, and start the viewer

sudo ./bin/k4aviewer

be careful! sudo must be added, or the permission is not enough

Write the serial port model of the camera into the rule script, so that the permission is higher and sudo free

sudo cp    /Azure-Kinect-Sensor-SDK/scripts/99-k4a.rules /etc/udev/rules.d/

Install ROS melody version of azure kinect

Key references: Robot capture (III) -- Installation of Azure Kinect SDK and ROS driver

Install opencv3 four point three

Download opencv3.0 from the official website four point three

cd ~/opencv-3.4.13
mkdir build 
cd build
cmake ..
build -j6
sudo make install

Install Azure_Kinect_ROS_Driver package

You don't need to transplant / ext files, as the old online tutorial said

# Download Azure_Kinect_ROS_Driver source code
cd catkin_ws/src
git clone

# Compile workspace
cd  catkin_ws


During compilation, you may encounter the problem of displaying the lack of cv::bridge. This is because xavier comes with opencv4, while ROS melody comes with cv::bridge, which is linked to the opencv directory and cannot be found

terms of settlement:

Method 1: modify cv_bridge configcmake path, making it compatible with the current opencv version
Key reference Blogs:
Solve CV in ROS Kinetic in Ubuntu environment_ Bridge and install opencv 2.4 XX compatibility notes

Method 2 (recommended): Download and compile your own CV in the workspace_ bridge
Reference blog:
NX Getting Started tutorial software - installing vins mono

cd  catkin_ws/src
git clone -b melodic
source devel/setup.bash
roslaunch azure_kinect_ros_driver driver.launch

Open rviz image topic to subscribe to images

Project Download

In order to facilitate rapid development, you can download the configured ros project and library files at the following link

There are three files
Will Azure_Kinect_ROS_Driver and vision_ The two function packages of OpenCV are put into their own catkin_ws/src workspace, compile libdepthangine so. 2.0 execution instructions

sudo cp /usr/lib/aarch64-linux-gnu

Added by grevathi_02 on Sat, 19 Feb 2022 00:28:51 +0200