Modular ROS 2 and YARP end-to-end pipeline for 3D scene reconstruction using manipulator arms

Abstract

We present an end-to-end pipeline for live 3D scene reconstruction, from raw point cloud acquisition to final mesh generation, ready for 3D printing. The hardware setup relies on a consumer-grade RGB-D camera (e.g., the Intel RealSense), an optional turntable controlled via a low-cost microcontroller (Arduino board), and a robotic arm (we used the ABB CRB 15000 “GoFa” collaborative robot). The underlying software architecture relies on fully open-source tools, with the OpenCV implementation of the KinectFusion algorithm at its backbone. This approach leverages the popular ROS 2 and YARP robotic middlewares, enabling a distributed network of reusable, modular components. We seek to provide a flexible framework that can easily be extended to a wider range of tools and robots. The system has been validated across three real scenarios; the 3D scanned meshes of three different objects (rubber duck, toy robot, human head) of varying sizes and features have been obtained, successfully generating 1:1 scale 3D-printed models. The source code of all developed components is openly available.

Publication
In 2026 International Conference on Automation, Control and Robotics (ICACR)
Accepted for IEEE ICACR 2026 (Nanjing, China, October 2026). The accepted manuscript is available as a PDF on this page; the DOI and official proceedings link will be added when they become available.
Qiang Li
Qiang Li
Professor and header of AG

My research interests include underwater robot, collaborative robots, humanoid robots.