README.md

Robotec GPU Lidar

About the project

Robotec GPU Lidar (RGL) is a cross-platform (Windows and Linux) C/C++ library developed by Robotec.AI for simulating LiDARs on CUDA-enabled GPUs, accelerated by RTX cores if available.

One of the use cases of RGL is implementing Lidar sensors in simulation engines. We are working on integrations with popular game/simulation engines:

• Unity
• O3DE
• Gazebo

If you would like to have a custom integration, feel free to contact us.

Features

Configurable LiDAR pattern and range	High performance
GPU-accelerated point cloud processing	Flexible pipeline creation

And more:

• Asynchronous raytracing
• Removing non-hit points
• Converting to custom binary output
• Downsampling *
• Writing to PCD file *
• Visualization *
• ROS2 publishing *
• Raw lidar packets publishing via UDP *
• Gaussian noise (see documentation)
• Instance/semantic segmentation
• Intensity based on provided textures
• Simple velocity distortion (only sensor velocity considered)

* extension required.

Runtime requirements

Hardware	Requirement
GPU	CUDA-enabled

Software	Requirement
Nvidia Driver (Linux)	>=515.43.04
Nvidia Driver (Windows)	>=472.50

Usage

An introduction to the RGL API along with an example can be found here.

Extensions

RobotecGPULidar library can be built with extensions enhancing RGL with additional functions:

• PCL - adds nodes and functions for point cloud processing that uses Point Cloud Library. See documentation.
• ROS2 - adds a node to publish point cloud messages to ROS2. Check ROS2 extension doc for more information, build instructions, and usage.
• UDP - adds a node to publish raw lidar packets, as emitted by physical lidar. Only available in the closed-source version.

Building in Docker (Linux)

1. Set up NVIDIA Container Toolkit
2. Download NVidia OptiX 7.2
3. export OptiX_INSTALL_DIR=<Path to OptiX>
4. docker build . --tag rgl:latest
5. docker run --net=host --gpus all -v $(pwd):/code -v ${OptiX_INSTALL_DIR}:/optix -e OptiX_INSTALL_DIR=/optix -e NVIDIA_DRIVER_CAPABILITIES=all -it rgl:latest /bin/bash
6. ./setup.py --clean-build --with-pcl
   • For build with ROS2 extension, do source ROS2 first:
     source /opt/ros/humble/setup.bash && ./setup.py --clean-build --with-pcl --with-ros2

Building on Ubuntu 22

1. Install CUDA Toolkit 11.7+.
2. Download NVidia OptiX 7.2.
   1. You may be asked to create a Nvidia account to download
3. Export environment variable:
   1. export OptiX_INSTALL_DIR=<your-OptiX-path>.
4. Use setup.py script to build.
   • It will use CMake to generate files for the build system (make) and the build.
   • You can pass optional CMake and make parameters, e.g.
     • ./setup.py --cmake="-DCMAKE_BUILD_TYPE=Debug" --make="-j 16"
   • You can build with extensions, e.g.
     • ./setup.py --with-pcl --with-ros2
   • See ./setup.py --help for usage information.

Building on Windows

1. Install Microsoft Visual Studio (Visual Studio 2019 when using ROS2 extension) with C++ CMake tools for Windows component.
2. Install CUDA Toolkit 11.4.4+.
3. Download NVidia OptiX 7.2.
   • install the framework and set the environment variable OptiX_INSTALL_DIR
4. Install Python3.
5. Run x64 Native Tools Command Prompt for VS 20xx and navigate to the RGL repository.
6. Run python setup.py command to build the project.
   • It will use CMake to generate files for the build system (ninja) and build.
   • You can pass optional CMake and ninja parameters, e.g.
     • python setup.py --cmake="-DCMAKE_BUILD_TYPE=Debug" --ninja="-j 16"
   • You can build with extensions, e.g.
     • ./setup.py --with-pcl --with-ros2
   • See python setup.py --help for usage information.

Acknowledgements

The development of this project was made possible thanks to cooperation with Tier IV - challenging needs in terms of features and performance of Tier IV's project allowed to significantly enrich Robotec GPU Lidar with features such as Gaussian noise and animated meshes as well as optimize it to provide real-time performance with many lidars.