11. Odin1 Relocalization Map Acquisition Guide

📄 Download English PDF: Odin1 Relocalization Map Acquisition Guide

11.1 Overview

Odin1 supports relocalization and is integrated with the MANIFOLD handheld scanner data pipeline. Three workflows are currently provided:

Option 1: Capture olx data with Odin, process it with Mindcloud, and use the Export Map function to export a bin file for Odin relocalization.
Option 2: Capture lx data with the MANIFOLD handheld scanner (e.g. Q9000), process it with Mindcloud, and use the Export Map function to export a bin file for Odin relocalization.
Option 3: Generate a bin file directly with Odin in SLAM mode for Odin relocalization.

11.2 How to Use

11.2.1 Option 1: Capture `olx` data with Odin and export the map

Record olx files with Odin (requires Odin firmware ≥ 0.10.0 and driver ≥ 0.9.0):

On Ubuntu, set recorddata: 1 in the driver to record data.
On Windows, run OdinViewer to record data.
Process the olx file with Mindcloud, then click Export Map in post-processing to obtain the relocalization bin file.

11.2.2 Option 2: Capture `lx` data with the MANIFOLD handheld scanner and export the map

(Skip this option if you do not have a MANIFOLD handheld scanner.)

Record lx files with the handheld scanner:

Connect the Q9000 to your phone and use Mindcloud Go to scan.
After scanning, export the project file to your local machine.
Process the lx file with Mindcloud, then click Export Map in post-processing to obtain the relocalization bin file.

11.2.3 Option 3: Generate a `bin` file with Odin SLAM mode

Record SLAM data with Odin:

On Ubuntu, set recorddata: 0 in the driver and set map mode to 1.
After recording, do not kill the driver process. Instead, open another terminal, go to /catkin_ws/src/odin_ros_driver, and run ./set_param.sh save_map 1. Wait until the driver terminal indicates that the map has been saved and a bin file has appeared in the corresponding folder before stopping.
Set map mode: 2 to switch to relocalization mode.
Re-run the driver to start relocalization.

11.3 Visualizing the Relocalization Map

Once the olx file is imported into Mindcloud, you can not only export the relocalization map but also click Save in the upper-left corner to save a point-cloud map in pcd format. This map can be displayed in RViz; you need to publish it as a ROS topic. You can downsample the pcd and write a custom publisher node — see the following Python sample:

pcd_publisher.py

The final result looks like this:

RViz display

11.4 Convert the relocalization `bin` directly to PLY with `map_to_ply` (no Mindcloud required)

If you just want to quickly visualize an existing relocalization map (loop_map.bin) and would rather not go through the full Mindcloud post-processing pipeline again, you can use the map_to_ply tool we provide to convert the bin file directly into a PLY point cloud, and then open it with common tools such as CloudCompare, MeshLab, or Open3D.

This tool works with any loop_map.bin (MAPV0001 format) produced by options 11.2.1 / 11.2.2 / 11.2.3.

11.4.1 Download

Pick the build that matches the CPU architecture of the machine where you’ll run the tool:

File	Target platform
map_to_ply_amd64	x86_64 (Intel / AMD)
map_to_ply_arm64	ARM aarch64 (Jetson / Kunpeng, etc.)

Built on Ubuntu 20.04 (glibc 2.31); compatible with Ubuntu 20.04 / 22.04 / 24.04.

11.4.2 Usage

After downloading, make it executable:

chmod +x map_to_ply_amd64    # or map_to_ply_arm64

Basic usage:

./map_to_ply_amd64 <input.bin> [output.ply]

input.bin — the input loop_map.bin
output.ply — the output PLY file (optional; defaults to the input name with a .ply extension)

Examples:

# Auto-name the output; produces loop_map.ply
./map_to_ply_amd64 loop_map.bin

# Specify the output name
./map_to_ply_amd64 loop_map.bin my_map.ply

11.4.3 Output

By default the tool writes a binary little-endian PLY containing each point’s x / y / z coordinates (float32). The resulting .ply can be opened directly in:

CloudCompare (recommended — cross-platform, handles large clouds well)
MeshLab
Open3D (Python: o3d.io.read_point_cloud("loop_map.ply"))
Blender (Import → PLY)

← Back to Odin1 (English)