Description

Sample code provided for working with Ouster sensors

To get started building the client and visualizer libraries, see the Sample Client and Visualizer section below. For instructions on ROS, start with the Example ROS Code section. Python SDK users should proceed straight to our python SDK homepage.

This repository contains sample code for connecting to and configuring ouster sensors, reading and visualizing data, and interfacing with ROS.

• ouster_client contains an example C++ client for ouster sensors
• ouster_viz contains a basic point cloud visualizer
• ouster_ros contains example ROS nodes for publishing point cloud messages
• python contains the code for the ouster sensor python SDK

Building the example code requires a compiler supporting C++11 and CMake 3.1 or newer and the tclap, jsoncpp, and Eigen3 libraries with headers installed on the system. The sample visualizer also requires the GLFW3 and GLEW libraries.

To install build dependencies on Ubuntu, run:

sudo apt install build-essential cmake libglfw3-dev libglew-dev libeigen3-dev \
     libjsoncpp-dev libtclap-dev

On macOS, install XCode and homebrew and run:

brew install cmake pkg-config glfw glew eigen jsoncpp tclap

To build run the following commands:

mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release <path to ouster_example>
make

where <path to ouster_example> is the location of the ouster_example source directory. The CMake build script supports several optional flags:

-DBUILD_VIZ=OFF                      Do not build the sample visualizer
-DBUILD_PCAP=ON                      Build pcap tools. Requres libpcap and libtins dev packages
-DBUILD_SHARED_LIBS                  Build shared libraries (.dylib or .so)
-DCMAKE_POSITION_INDEPENDENT_CODE    Standard flag for position independent code

The example code can be built on Windows 10 with Visual Studio 2019 using CMake support and vcpkg for dependencies. Follow the official documentation to set up your build environment:

• Visual Studio
• Visual Studio CMake Support
• Visual Studio CPP Support
• Vcpkg, at tag "2020.11-1" installed and integrated with Visual Studio

Note You'll need to run git checkout 2020.11-1 in the vcpkg directory before bootstrapping to use the correct versions of the dependencies. Building may fail unexpectedly if you skip this step.

Don't forget to integrate vcpkg with Visual Studio after bootstrapping:

.\vcpkg.exe integrate install

You should be able to install dependencies with:

.\vcpkg.exe install --triplet x64-windows glfw3 glew tclap jsoncpp eigen3

After these steps are complete, you should be able to open, build and run the ouster_example project using Visual Studio:

1. Start Visual Studio.
2. When the prompt opens asking you what type of project to open click Open a local folder and navigate to the ouster_example source directory.
3. After opening the project for the first time, wait for CMake configuration to complete.
4. Make sure Visual Studio is building in release mode. You may experience performance issues and missing data in the visualizer otherwise.
5. In the menu bar at the top of the screen, select Build > Build All.

6. To use the resulting binaries, go to View > Terminal and run, for example:

   .\out\build\x64-Release\ouster_client\ouster_client_example.exe -h

Make sure the sensor is connected to the network. See "Connecting to the Sensor" in the Software User Manual for instructions and different options for network configuration.

Navigate to ouster_client under the build directory, which should contain an executable named ouster_client_example. This program will attempt to connect to the sensor, capture lidar data, and write point clouds out to CSV files:

./ouster_client_example <sensor hostname> <udp data destination>

where <sensor hostname> can be the hostname (os-99xxxxxxxxxx) or IP of the sensor and <udp data destingation> is the hostname or IP to which the sensor should send lidar data.

On Windows, you may need to allow the client/visualizer through the Windows firewall to receive sensor data.

Navigate to ouster_viz under the build directory, which should contain an executable named simple_viz . Run:

./simple_viz <flags> <sensor hostname> <udp data destination>

where <sensor hostname> can be the hostname (os-99xxxxxxxxxx) or IP of the sensor and <udp data destingation> is the hostname or IP to which the sensor should send lidar data.

The sample visualizer does not currently include a GUI, but can be controlled with the mouse and keyboard:

• Click and drag rotates the view
• Middle click and drag moves the view
• Scroll adjusts how far away the camera is from the vehicle

Keyboard controls:

key	what it does
p	Increase point size
o	Decrease point size
m	Cycle point cloud coloring mode
v	Toggle range cycling
n	Toggle display near-IR image from the sensor
shift + r	Reset camera
e	Change size of displayed 2D images
;	Increase spacing in range markers
'	Decrease spacing in range markers
r	Toggle auto rotate
w	Camera pitch up
s	Camera pitch down
a	Camera yaw left
d	Camera yaw right
1	Toggle point cloud visibility
0	Toggle orthographic camera
=	Zoom in
-	Zoom out
shift	Camera Translation with mouse drag

For usage and other options, run ./simple_viz -h

The sample code include tools for publishing sensor data as standard ROS topics. Since ROS uses its own build system, it must be compiled separately from the rest of the sample code.

The provided ROS code has been tested on ROS Kinetic, Melodic, and Noetic on Ubuntu 16.04, 18.04, and 20.04, respectively. Use the installation instructions to get started with ROS on your platform.

The build dependencies include those of the sample code:

sudo apt install build-essential cmake libglfw3-dev libglew-dev libeigen3-dev \
     libjsoncpp-dev libtclap-dev

Additionally, you should install the ros dependencies:

sudo apt install ros-<ROS-VERSION>-ros-core ros-<ROS-VERSION>-pcl-ros \
     ros-<ROS-VERSION>-tf2-geometry-msgs ros-<ROS-VERSION>-rviz

where <ROS-VERSION> is kinetic, melodic, or noetic.

Alternatively, if you would like to install dependencies with `rosdep`:

rosdep install --from-paths <path to ouster example>

To build:

source /opt/ros/<ROS-VERSION>/setup.bash
mkdir -p ./myworkspace/src
cd myworkspace
ln -s <path to ouster_example> ./src/
catkin_make -DCMAKE_BUILD_TYPE=Release

Warning: Do not create your workspace directory inside the cloned ouster_example repository, as this will confuse the ROS build system.

For each command in the following sections, make sure to first set up the ROS environment in each new terminal by running:

source myworkspace/devel/setup.bash

Make sure the sensor is connected to the network. See "Connecting to the Sensor" in the Software User Manual for instructions and different options for network configuration.

To publish ROS topics from a running sensor, run:

roslaunch ouster_ros ouster.launch sensor_hostname:=<sensor hostname> \
                                   udp_dest:=<udp data destination> \
                                   metadata:=<path to metadata json> \
                                   lidar_mode:=<lidar mode> viz:=<viz>

where:

• <sensor hostname> can be the hostname (os-99xxxxxxxxxx) or IP of the sensor
• <udp data destination> is the hostname or IP to which the sensor should send data
• <path to metadata json> is the path to the json file to which to save calibration metadata
• <lidar mode> is one of 512x10, 512x20, 1024x10, 1024x20, or 2048x10, and
• <viz> is either true or false: if true, a window should open and start displaying data after a few seconds.

Note that by default the working directory of all ROS nodes is set to ${ROS_HOME}, generally $HOME/.ros, so if metadata is a relative path, it will write to that path inside ${ROS_HOME}. To avoid this, you can provide an absolute path to metadata.

To record raw sensor output use rosbag record. After starting the roslaunch command above, in another terminal, run:

rosbag record /os_node/imu_packets /os_node/lidar_packets

This will save a bag file of recorded data in the current working directory.

You should copy and save the metadata file alongside your data. The metadata file will be saved at the provided path to roslaunch. If you run the node and cannot find the metadata file, try looking inside your ${ROS_HOME}, generally $HOME/.ros. Regardless, you must retain the metadata file, as you will not be able to replay your data later without it.

To publish ROS topics from recorded data, specify the replay and metadata parameters when running roslaunch:

roslaunch ouster_ros ouster.launch replay:=true metadata:=<path to metadata json>

And in a second terminal run rosbag play:

rosbag play --clock <bag files ...>

A metadata file is mandatory for replay of data. See Recording Data for how to obtain the metadata file when recording your data.

Python SDK users should proceed straight to the Ouster python SDK homepage.

• Sample sensor output usable with the provided ROS code is available here.
• For network configuration, refer to "Connecting to the Sensor" in the Software User Manual.