Hint

Please change branch to Bunker-DVI-Dataset-reg-1 for quick experiment.

Example Dataset:

Download the dataset from Bunker DVI Dataset

benchmark-EllipseLIO-to-HDMapping

Runs the EllipseLIO LiDAR-Inertial odometry algorithm on a ROS 2 bag file and converts the output to an HDMapping session.

EllipseLIO is an Adaptive LiDAR Inertial Odometry approach with an Ellipsoid Representation, by V4RL, ETH/UCY.

Prerequisites

• Docker
• A ROS 2 bag containing a sensor_msgs/msg/PointCloud2 topic and a sensor_msgs/msg/Imu topic matching the topic names declared in the chosen EllipseLIO config file (ROS 1 bags are automatically converted to ROS 2 format).

Step 1 — Clone with submodules

git clone https://github.com/MapsHD/benchmark-EllipseLIO-to-HDMapping.git --recursive
cd benchmark-EllipseLIO-to-HDMapping

Step 2 — Build the Docker image

docker build -t ellipselio_humble .

This installs:

• Ubuntu 22.04 + ROS 2 Humble
• Eigen3, PCL, OpenCV, Boost, OpenMP
• EllipseLIO (compiled from submodule)
• ROS 2 workspace with ellipselio and ellipselio-to-hdmapping

The build takes several minutes on first run.

Step 3 — Run the pipeline

chmod +x docker_session_run-ros2-ellipselio.sh
./docker_session_run-ros2-ellipselio.sh /path/to/input.bag /path/to/output/dir

Or with no arguments to use a GUI file selector (requires zenity):

./docker_session_run-ros2-ellipselio.sh

By default the script uses the mid360_hdmapping.yaml EllipseLIO config, the Livox Mid-360 config the EllipseLIO authors prepared for HDMapping (Bunker DVI dataset). Pick a different one with the CONFIG_FILE environment variable, e.g.:

CONFIG_FILE=os64_ncd.yaml ./docker_session_run-ros2-ellipselio.sh /path/to/input.bag /path/to/output/dir

Available configs (from the ellipselio/config/ directory):

Config	LiDAR	Typical dataset
mid360_hdmapping.yaml	Livox Mid-360	Bunker DVI (HDMapping) — default
mid360_grandtour.yaml	Livox Mid-360	GrandTour
avia_geode.yaml	Livox Avia	GEODE Gamma
qt64_spires.yaml	Hesai Pandar QT64	Oxford Spires
xt32_grandtour.yaml	Hesai XT-32	GrandTour
os64_ncd.yaml	Ouster OS-64	Newer College
os128_ncd.yaml	Ouster OS-128	Newer College
os64_geode.yaml	Ouster OS-64	GEODE Beta
vlp16_bot.yaml	Velodyne VLP-16	BotanicGarden
vlp16_geode.yaml	Velodyne VLP-16	GEODE Alpha
vlp16_graco.yaml	Velodyne VLP-16	GraCo
vlp16_grandtour.yaml	Velodyne VLP-16	GrandTour

This repository's config/ directory adds livox_reg.yaml (the earlier benchmark config for the Bunker DVI Mid-360). A file in config/ is mounted into the container and takes precedence over a package config of the same name.

What happens:

The script opens a Docker container with a tmux session containing five panes (six with a Livox config) and a control window:

Pane	Role
0	ros2 launch ellipselio ellipselio_standalone.launch.py — subscribes to the LiDAR + IMU topics from the chosen config, publishes /ellipselio_odom + /cloud_scan
1	RViz2 — live visualization of the EllipseLIO map and trajectory
2	ros2 bag record — captures the two published topics
3	ros2 bag play — plays your input bag with simulated clock
4	diagnostics — shows active topics and publishing rates
5	Livox configs only (lidar.type: 1): livox_format_bridge — converts the bag's /livox/pointcloud (livox_ros_driver layout, per-point float time offset) into EllipseLIO's Livox point format (absolute double timestamp) and publishes it on the config's lidar.topic
control	auto-shutdown — waits for playback to finish, then stops all nodes

After playback completes, the control window automatically stops the recorder, kills all nodes, and exits tmux. A second Docker run then converts the recorded bag into the HDMapping session format.

Step 4 — Open in HDMapping

Output files appear in <output_dir>/output_hdmapping-EllipseLIO/:

lio_initial_poses.reg
poses.reg
scan_lio_0.laz
scan_lio_1.laz
...
session.json
trajectory_lio_0.csv
trajectory_lio_1.csv
...

Open session.json with the multi_view_tls_registration_step_2 application.

Notes on EllipseLIO

EllipseLIO requires both LiDAR point clouds and IMU data. The input topic names are configured inside the YAML config file (not as command-line parameters like D-LIO). The relevant fields are:

lidar:
  type: 1|2|3|4   # LIVOX, VELODYNE, OUSTER, HESAI
  rate: 10
  topic: "/your/lidar/topic"
  t_imu_lidar: [x, y, z]
  r_imu_lidar: [qx, qy, qz, qw]   # or a 3x3 rotation matrix (9 values, row-major)
imu:
  rate: 200
  topic: "/your/imu/topic"

If the topic names in your bag differ from the ones in the chosen config, either pick another config or put a customised config into this repository's config/ directory and select it with CONFIG_FILE=<file name>.

The recorded topics (used by the converter) are also tunable via env vars:

Variable	Meaning	Default
ODOM_TOPIC	EllipseLIO odometry output	/ellipselio_odom
CLOUD_TOPIC	EllipseLIO per-scan cloud (world frame)	/cloud_scan

Note: EllipseLIO publishes /cloud_scan already expressed in the odom_ellipselio (world) frame, so the converter does not re-apply the odometry to the points — it only uses /ellipselio_odom to build the per-chunk trajectory files.

Contact

januszbedkowski@gmail.com