Nebula Sensor Driver

Nebula is a sensor driver platform that is designed to provide a unified framework for as wide a variety of devices as possible. While it primarily targets Ethernet-based LiDAR sensors, it aims to be easily extendable to support new sensors and interfaces. Nebula provides the following features:

• Support for Velodyne and Hesai sensors, with other LiDAR vendor support under development
• ROS 2 interface implementations
• TCP/IP and UDP communication implementations
• Abstraction of sensor decoders and hardware interfaces available as libraries
• Handling of standard LiDAR functionality, including but not limited to:
  • Configuration of communication settings such as sensor and host IP addresses and communication ports
  • Configuration of scan speed, synchronization settings, scan phase, and field of view
  • Receiving and conversion of UDP packet data into point clouds in Cartesian co-ordinates
  • Receiving and interpretation of diagnostics information from the sensor
  • Support for multiple return modes and labelling of return types for each point

With a rapidly increasing number of sensor types and models becoming available, and varying levels of vendor and third-party driver support, Nebula creates a centralized driver methodology. We hope that this project will be used to facilitate active collaboration and efficiency in development projects by providing a platform that reduces the need to re-implement and maintain many different sensor drivers. Contributions to extend the supported devices and features of Nebula are always welcome.

How to build

Nebula builds on ROS Galactic and Humble.

Note

A TCP enabled version of ROS' Transport Driver is required to use Nebula. It is installed automatically into your workspace using the below commands. However, if you already have ROS transport driver binaries installed, you will have to uninstall them to avoid conflicts (replace humble with your ROS distribution): sudo apt remove ros-humble-udp-driver ros-humble-io-context

To build Nebula run the following commands in your workspace:

# In workspace
mkdir src
git clone https://github.com/tier4/nebula.git src
# Import dependencies
vcs import src < src/build_depends.repos
rosdep install --from-paths src --ignore-src -y -r
# Build Nebula
colcon build --symlink-install --cmake-args -DCMAKE_BUILD_TYPE=Release

How to run tests

Run tests:

colcon test --event-handlers console_cohesion+ --packages-above nebula_common

Show results:

colcon test-result --all

Generic launch file

You can easily run the sensor hardware interface, the sensor hardware monitor and sensor driver using (e.g. Pandar64):

ros2 launch nebula_ros nebula_launch.py sensor_model:=Pandar64

If you don't want to launch the hardware (i.e. when you are working from a rosbag), set the launch_hw flag to false:

ros2 launch nebula_ros nebula_launch.py sensor_model:=Pandar64 launch_hw:=false

If you don't want the hardware driver to perform the sensor configuration communication (i.e. limited number of connections) set the setup_sensor flag to false:

ros2 launch nebula_ros nebula_launch.py sensor_model:=Pandar64 setup_sensor:=false

You should ideally provide a config file for your specific sensor, but default ones are provided nebula_drivers/config:

ros2 launch nebula_ros nebula_launch.py sensor_model:=Pandar64 config_file:=your_sensor.yaml

Supported sensors

Supported models, where sensor_model is the ROS param to be used at launch:

Manufacturer	Model	sensor_model	Configuration file	Test status
HESAI	Pandar 64	Pandar64	Pandar64.yaml	✔️
HESAI	Pandar 40P	Pandar40P	Pandar40P.yaml	✔️
HESAI	Pandar XT32	PandarXT32	PandarXT32.yaml	✔️
HESAI	Pandar XT32M	PandarXT32M	PandarXT32M.yaml	⚠️
HESAI	Pandar QT64	PandarQT64	PandarQT64.yaml	✔️
HESAI	Pandar QT128	PandarQT128	PandarQT128.yaml	⚠️
HESAI	Pandar AT128	PandarAT128	PandarAT128.yaml	✔️*
HESAI	Pandar 128E4X	Pandar128E4X	Pandar128E4X.yaml	⚠️
Velodyne	VLP-16	VLP16	VLP16.yaml	⚠️
Velodyne	VLP-16-HiRes	VLP16		❌
Velodyne	VLP-32	VLP32	VLP32.yaml	⚠️
Velodyne	VLS-128	VLS128	VLS128.yaml	⚠️

Test status:
✔️: complete
⚠️: some functionality yet to be tested
❌: untested
*: AT128 needs software version 3.50.8 or newer for the scan_angle setting to work correctly.

ROS parameters

Common ROS parameters

Parameters shared by all supported models:

Parameter	Type	Default	Accepted values	Description
sensor_model	string		See supported models
return_mode	string		See supported return modes
frame_id	string	Sensor dependent		ROS frame ID
scan_phase	double	0.0	degrees [0.0, 360.0]	Scan start angle

Hesai specific parameters

Supported return modes per model

Sensor model	return_mode	Mode
Pandar XT32M	Last	Single
Pandar XT32M	Strongest	Single
Pandar XT32M	LastStrongest	Dual
Pandar XT32M	First	Single
Pandar XT32M	LastFirst	Dual
Pandar XT32M	FirstStrongest	Dual
Pandar XT32M	Dual	Dual
---	---	---
Pandar AT128	Last	Single
Pandar AT128	Strongest	Single
Pandar AT128	LastStrongest	Dual
Pandar AT128	First	Single
Pandar AT128	LastFirst	Dual
Pandar AT128	FirstStrongest	Dual
Pandar AT128	Dual	Dual
---	---	---
Pandar QT128	Last	Single
Pandar QT128	Strongest	Single
Pandar QT128	LastStrongest	Dual
Pandar QT128	First	Single
Pandar QT128	LastFirst	Dual
Pandar QT128	FirstStrongest	Dual
Pandar QT128	Dual	Dual
---	---	---
Pandar QT64	Last	Single
Pandar QT64	Dual	Dual
Pandar QT64	First	Single
---	---	---
Pandar 40P	Last	Single
Pandar 40P	Strongest	Single
Pandar 40P	Dual	Dual
---	---	---
Pandar 64	Last	Single
Pandar 64	Strongest	Single
Pandar 64	Dual	Dual

Hardware interface parameters

Parameter	Type	Default	Accepted values	Description
frame_id	string	hesai		ROS frame ID
sensor_ip	string	192.168.1.201		Sensor IP
host_ip	string	255.255.255.255		Host IP
data_port	uint16	2368		Sensor port
gnss_port	uint16	2369		GNSS port
frequency_ms	uint16	100	milliseconds, > 0	Time per scan
packet_mtu_size	uint16	1500		Packet MTU size
rotation_speed	uint16	600		Rotation speed
cloud_min_angle	uint16	0	degrees [0, 360]	FoV start angle
cloud_max_angle	uint16	359	degrees [0, 360]	FoV end angle
dual_return_distance_threshold	double	0.1		Dual return distance threshold
diag_span	uint16	1000	milliseconds, > 0	Diagnostic span
setup_sensor	bool	True	True, False	Configure sensor settings

Driver parameters

Parameter	Type	Default	Accepted values	Description
frame_id	string	hesai		ROS frame ID
calibration_file	string			LiDAR calibration file
correction_file	string			LiDAR correction file

Velodyne specific parameters

Supported return modes

return_mode	Mode
SingleFirst	Single (First)
SingleStrongest	Single (Strongest)
SingleLast	Single (Last)
Dual	Dual

Hardware interface parameters

Parameter	Type	Default	Accepted values	Description
frame_id	string	velodyne		ROS frame ID
sensor_ip	string	192.168.1.201		Sensor IP
host_ip	string	255.255.255.255		Host IP
data_port	uint16	2368		Sensor port
gnss_port	uint16	2369		GNSS port
frequency_ms	uint16	100	milliseconds, > 0	Time per scan
packet_mtu_size	uint16	1500		Packet MTU size
cloud_min_angle	uint16	0	degrees [0, 360]	FoV start angle
cloud_max_angle	uint16	359	degrees [0, 360]	FoV end angle

Driver parameters

Parameter	Type	Default	Accepted values	Description
frame_id	string	velodyne		ROS frame ID
calibration_file	string			LiDAR calibration file
min_range	double	0.3	meters, >= 0.3	Minimum point range published
max_range	double	300.0	meters, <= 300.0	Maximum point range published
cloud_min_angle	uint16	0	degrees [0, 360]	FoV start angle
cloud_max_angle	uint16	359	degrees [0, 360]	FoV end angle

Software design overview

How to evaluate performance

You can evaluate Nebula performance on a given rosbag and sensor model using the below tools. The profiling runner is most accurate when assigning isolated cores via the -c <core_id>. CPU frequencies are locked/unlocked automatically by the runner to increase repeatability.

Run profiling for each version you want to compare:

./scripts/profiling_runner.bash baseline -m Pandar64 -b ~/my_rosbag -c 2 -t 20 -n 3
git checkout my_improved_branch
./scripts/profiling_runner.bash improved -m Pandar64 -b ~/my_rosbag -c 2 -t 20 -n 3

Show results:

pip3 install scripts/requirements.txt  # first-time setup
python3 scripts/plot_times.py baseline improved