Students Final Project Presentation - Robotics Developer Masterclass

Why It Matters

By proving that precise, low‑cost perception and reliable motion planning can automate routine service tasks, the project paves the way for scalable commercial coffee‑robot solutions and enriches hands‑on robotics education.

Key Takeaways

• OpenCV circle detection replaced YOLO for cup‑holder perception.
• UR3 arm with 0.5 mm positional tolerance ensures upright cup placement.
• Foxglove dashboard enables real‑time ordering and status monitoring.
• Behavior tree handles detection, placement, and graceful recovery failures.
• System integrates RGB‑D camera, MoveIt2 planning, and custom ROS2 messages.

Summary

The video showcases Matias Rosas’s final project for the Robotics Developer Masterclass: an autonomous coffee‑cup dispenser built around a Universal Robotics UR3 arm, a static tray holder, and a web‑based control interface. Drawing on his data‑science and mechatronics background, he created a service‑robot prototype that can detect empty cup slots, pick up a cup, and place it upright without spilling. Key technical insights include a shift from heavyweight YOLO models to lightweight OpenCV circle detection, leveraging depth checks to differentiate shadows from occupied slots. Motion planning is handled by MoveIt2 with OMPL, enforcing 0.5 mm positional and 0.04 rad orientation tolerances, and limiting speed to 10 % of the arm’s maximum to avoid coffee spillage. A ROS2‑based behavior tree orchestrates perception, manipulation, and graceful recovery when placement fails. Notable implementation details feature a custom TF‑based message pipeline between detection and manipulation, a Foxglove dashboard that visualizes point clouds and lets users order cup placement in real time, and collision objects modeled as simple blocks to simplify safety checks. The system also includes a fallback routine that returns the cup to its original position and prompts the user to select another slot. The project demonstrates how modular, low‑cost perception and robust planning can make service robots viable for repetitive tasks like coffee dispensing. Its open‑source ROS2 stack offers a replicable blueprint for future deployments in cafeterias, labs, or other hospitality settings.

Original Description

Presentations of the Final Projects of the Robotics Developer Masterclass (https://bit.ly/49OBvyn).

These presentations are open to the public and provide an opportunity to learn about the different research topics developed by the students of this program.

The Masterclass Final Project consists of the student undertaking an autonomous project under the direction of a tutor, the completion of which is intended to help the students implement the robotics development skills that they have gained during their studies.

Agenda:

16:00 CEST - Mathias Rosas - Automated Coffee Dispenser

Tutor(s): Alberto Ezquerro

--

#ros #robotics #robot #ai #ros2