Link to LinkedIn post with the video
Welcome to the Open-Source 3D Mapping System project! This repository contains everything you need to build your own 3D mapping system, including 3D-printable modular parts, sensor initialization scripts, and code to start SLAM (Simultaneous Localization and Mapping) systems. You'll also find code for data recording from the sensors.
🤖 Want to build your own 3D mapping system? You're in the right place!
I've just open-sourced the 3D-printable modular parts that allow you to build a versatile system for both robots and handheld applications. These components were originally designed for the Go1-legged robot, but can be adapted for use with any robot with a flat surface, or even used as a standalone handheld system.
You can explore and download the 3D-printable parts from the following link:
- Modular Design: Compatible with different configurations for robots or handheld systems.
- Supports:
- Ouster 3D LiDAR
- 4 Intel D435i RealSense cameras
- NUC mini PC
- Two batteries
- Automatic initialization of sensors and starting SLAM systems.
- Easy recording of data for research and analysis.
To get up and running with the system:
-
Download the 3D parts and assemble the hardware components according to your use case (robot or handheld).
-
Clone this repository:
git clone https://github.com/MigVeg/MappingSystem.git
-
Follow the setup guide to initialize the sensors and start the SLAM system, you can do that simply by runing the following command:
bash auto_run
-
If you plan to use the data later, it’s advisable to focus on recording only and avoid running the SLAM system simultaneously. The SLAM system is resource-intensive and can quickly discharge the batteries.
The modular mounting system allows easy attachment of sensors to your robot or for handheld use. It's originally designed for the Go1-legged robot, but its flexibility makes it usable for any robotic platform with a flat surface.
For details on how the components fit together, check out our research paper:
The code in this repository allows you to initialize the following sensors:
- Ouster 3D LiDAR
- 4 x Intel D435i RealSense cameras
Once the sensors are initialized, you can start the SLAM system and capture live data. You can also record this data for further research and analysis.
If you use this project or its 3D parts for academic research, please make sure to cite the following paper and resource:
@inbook{MountingSystem2024:vega:paper,
doi = {10.13154/294-10094},
url = {https://hss-opus.ub.ruhr-uni-bochum.de/opus4/10094},
author = {Vega-Torres, Miguel Arturo and Pfitzner, Fabian},
language = {en},
title = {Investigating robot dogs for construction monitoring},
publisher = {Ruhr-Universität Bochum},
year = {2023},
copyright = {Creative Commons - CC BY 4.0 - Namensnennung 4.0 International}
}
@misc{MountingSystem2024:vega:data,
author = {Vega-Torres, Miguel A. and Borrmann, André},
title = {{CMS Sensor Mounting System}},
year = {2024},
type = {Dataset},
abstract = {The data provided here includes the mounting system, which enable the mounting of multiple sensors on a robot. Although it was specifically developed for the Go1-legged robot, it can be utilized on any robot with a flat surface or as a handheld system by detaching the sensors from the components that hold the PC and batteries.
The mounting system is designed to accommodate an Ouster 3D LiDAR and four D435i RealSense cameras. Additionally, it supports a NUC mini PC and two batteries. For detailed information about the connections and usage of the system, please refer to this https://mediatum.ub.tum.de/node?id=1720803
Also, you can see the model in 3D: https://a360.co/46tjrIc},
keywords = {3D Mesh; Inventor; Mounting; LiDAR; Camera; Mapping; SLAM},
doi = {10.14459/2024mp1750434}
}
✨ Special thanks to Felix Eickeler for helping develop the first versions of this system. And to Fabian Pfitzner for the collaboration during the writing of the paper and recording of the sequences at the construction site! 👷♂️🏗
🌟 Also, thanks to:
- André Borrmann, Denis Wohlfeld, Alexander Braun, Martin Slepicka