Matteo Calabria
Master's student in Industrial Automation Engineering. Currently working on humanoid robotics, reinforcement learning, and VLMs/VLAs for embodied intelligence.
Experience
Research Assistant in Humanoid Robotics
CurrentWorking on humanoid robotics in the Embodied Mobile Agents (EMA) group at the Centre for Artificial Intelligence (CAI), School of Engineering.
Research Intern in RL for Humanoid Robotics
Trained reinforcement learning locomotion policies for quieter walking on the Unitree G1, learning a footstep vibration model from hardware and using it as an extra reward term during training.
Research Intern in ML for Anesthesia Control System
Designed and implemented a deep learning-driven control system within the ACTIVA Project to optimize intravenous anesthesia delivery, significantly advancing patient safety and anesthesia precision.
Education
Double Degree in Robotics and Autonomous Systems
2025 - 2026Interdisciplinary curriculum covering mechatronics, sensor fusion, reinforcement learning, and human-robot interaction.
M.Sc. in Industrial Automation Engineering
2024 - PresentSpecializing in robotics and control sytems applications for industrial automation.
B.Sc. in Industrial Automation Engineering
2021 - 2024Focused on robotics control and automation fundamentals.
Summer Schools
Imperial Robotics Summer School
Jul 2025Foundational and specialized lectures covering kinematics, robot learning, and control, plus a week-long project implementing an RL policy for the Unitree Go2 robot with multiple locomotion strategies.
PhD Summer School: Reinforcement Learning and LLMs for Robotics
Jul 2025Practice-oriented sessions on LLM fundamentals and RL algorithms, including HPC deployment workflows and CloudGripper robot control via LLM instructions.
Featured Projects
DreamMotion
Mar 2026HACK26Built a voice-to-robot-motion pipeline where natural language commands are processed by a GPT-4.1 agent, converted to motion sequences by NVIDIA's Kimodo diffusion model, and deployed to the Unitree G1 humanoid in simulation or on real hardware. Developed in 48 hours at HACK26.
ETH Robotics ClubHuman-Robot Interaction Playground
Dec 2025Developed an imitation learning framework implementing Behavioral Cloning, DAgger, and DAgger+Replay. Benchmarked against RL baselines (PPO, SAC) on MuJoCo environments.
Sorbonne UniversitéLearning Robot Manipulation from Demonstrations
Sep 2025Assembled and 3D printed two open-source SO-101 robotic arms, recorded synchronized kinematic and video data, and trained ACT and smolVLA policies on the collected dataset.
Reolink-DetectAI
Jun 2025Custom home security system using OpenCV and YOLOv8 to monitor RTSP camera streams, perform object detection, and send email alerts with detected images.
Downhill Event Detection
May 2025Developed a Machine Learning system to detect jumps and turns in downhill mountain biking using high-frequency IMU and fork sensor data. Took part in a field measurement campaign, collecting and processing 300k+ sensor samples.
University of BresciaAutomatic Control of Anesthesia
Sep 2024Bachelor's ThesisDesigned a deep learning-driven control system within the ACTIVA Project to predict BIS values and optimize intravenous anesthesia delivery, overcoming traditional closed-loop limitations.
University of BresciaCommunities
Member
CurrentCommunityETH Robotics Club
Student-led robotics community at ETH Zürich, enabling the next generation of robotics talent through projects, events, and industry partnerships.
Learn More →
Ambassador & Tech Team
CurrentMentorshipLead The Future Mentorship
Among the few Italian students selected to be mentees for LeadTheFuture, a leading mentorship non-profit for STEM students, with an acceptance rate below 15%.
Learn More →
Fellow
CurrentFellowshipEuroTech Federation
Federating the best builders of today to create the Europe of tomorrow
Learn More →Honors & Awards
ISSA PULIRE Hackathon 2025
🥇 1st PlaceAI-Orchestrator: A real-time web app for managing stadium cleaning operations using AI. Built in just 6 hours, combining computer vision, forecasting, and smart task allocation.
ABB RoboCup 2025
🥈 2nd PlaceParticipated with University of Brescia team in a national-level robotics competition. Programmed virtual robotic cells and deployed pick-and-place tasks on real industrial robots.
