GENZ-ICP converter to HDMapping

Hint

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

Intended use

This small toolset allows to integrate SLAM solution provided by genz-icp with HDMapping. This repository contains ROS 2 workspace that :

• submodule to tested revision of GenZ ICP
• a converter that listens to topics advertised from odometry node and save data in format compatible with HDMapping.

Dependencies

sudo apt install -y nlohmann-json3-dev

Building

Clone the repo

mkdir -p /test_ws/src
cd /test_ws/src
git clone https://github.com/marcinmatecki/GenZ-ICP-to-HDMapping.git --recursive
cd ..
colcon build

Usage - data SLAM:

Prepare recorded bag with estimated odometry:

In first terminal record bag:

ros2 bag record /genz/local_map /genz/odometry

start odometry:

cd /test_ws/
source ./install/setup.sh # adjust to used shell
ros2 launch genz_icp odometry.launch.py topic:=<topic_name>
ros2 bag play <rosbag_file_name>.db3

Usage - conversion:

cd /test_ws/
source ./install/setup.sh # adjust to used shell
ros2 run genz-icp-to-hdmapping listener <recorded_bag> <output_dir>

Example:

Download the dataset from NTU-VIRAL For this example, download eee_03.

Convert(If it's a ROS1 .bag file):

rosbags-convert --src {your_downloaded_bag} --dst {desired_destination_for_the_converted_bag}

Record the bag file:

ros2 bag record /genz/local_map /genz/odometry -o {your_directory_for_the_recorded_bag}

GenZ Launch:

cd /test_ws/
source ./install/setup.sh # adjust to used shell
ros2 launch genz_icp odometry.launch.py topic:=/os1_cloud_node1/points
ros2 bag play <rosbag_file_name>.db3

During the record (if you want to stop recording earlier) / after finishing the bag:

In the terminal where the ros record is, interrupt the recording by CTRL+C
Do it also in ros launch terminal by CTRL+C.

Usage - Conversion (ROS bag to HDMapping, after recording stops):

cd /test_ws/
source ./install/setup.sh # adjust to used shell
ros2 run genz-icp-to-hdmapping listener <recorded_bag> <output_dir>

Running with Docker

To run the complete pipeline using Docker:

1. Build the Docker image:

./build-docker.sh

2. Convert ROS1 bags to ROS2 format (if needed):

./run_docker_convert_to_ros2.sh

Note: Only run this step if your dataset contains ROS1 .bag files that need to be converted to ROS2 format.

After conversion:

data
└── ConSLAM
    ├── sequence1
    │   ├── converted
    │   │   ├── converted.db3
    │   │   └── metadata.yaml
    │   ├── Copy of recording.bag
    ├── sequence2
    │   ├── converted
    │   │   ├── converted.db3
    │   │   └── metadata.yaml
    │   ├── Copy of recording.bag

3. Process all sequences:

./run_all_sequences.sh

This script runs the GenZ-ICP odometry processing for all sequences using Docker Compose.

4. Convert results to HDMapping format:

./run_docker_convert_result_to_hdmapping.sh

This script converts the recorded GenZ-ICP results to HDMapping-compatible format for all processed sequences.

Results

data
└── ConSLAM
    ├── sequence1
    │   ├── converted
    │   │   ├── converted.db3
    │   │   └── metadata.yaml
    │   ├── Copy of recording.bag
    │   ├── results-genz-icp
    │   │   └── rosbag2_2025_10_20-21_08_46
    │   │       ├── hdmapping
    │   │       └── rosbag2_2025_10_20-21_08_46_0.mcap
    ├── sequence2
    │   ├── converted
    │   │   ├── converted.db3
    │   │   └── metadata.yaml
    │   ├── Copy of recording.bag
    │   ├── results-genz-icp
    │   │   └── rosbag2_2025_10_20-21_16_05
    │   │       ├── hdmapping
    │   │       └── rosbag2_2025_10_20-21_16_05_0.mcap