Hi!
I'm Gennaro Scarati
Robotics & AI Engineer
About Me
I am a Robotics and AI engineer passionate about creating systems that move, sense, and learn. My background spans robotics, artificial intelligence and control systems, with experience in both startups and large-scale production environments. I’d describe myself as a very curious person, always eager to learn new things and stay updated with the latest technologies. In my free time, I enjoy hiking and exploring new places on foot.
Technical skills
Languages
Experience
Robotics & AI Engineer
Eurecat
Barcelona, Spain
July 2023 - Present
July 2023 - Present
Barcelona, Spain
- •Develop and deploy robotics and AI software modules for industrial and agricultural robots.
- •Implemented modular behavior trees orchestrating high-level dexterous manipulation tasks, reducing development time by 50%, while improving trajectory optimality and feasibility.
- •Designed a real-time perception pipeline for detection, segmentation, tracking, 6D pose, and size estimation, achieving sub-centimeter accuracy for pose and size.
- •Train and implement learning-from-demonstration pipelines (VLAs, Diffusion Policies) to solve dexterous manipulation tasks, such as fruit harvesting and industrial assembly.
Control Systems Engineer
Dumarey Softronix
Turin, Italy
Mar 2022 - July 2023
Mar 2022 - July 2023
Turin, Italy
- •Developed and maintained control systems for General Motors vehicles, deployed on around 100,000 units in 2024 alone.
- •Enhanced fault diagnostic performance by 50% by combining RNN-based system prediction with classical model-based methods, while reducing calibration effort by 30%.
- •Collaborated with cross-functional teams in Italy, India, and the U.S. to meet development, calibration, and testing milestones.
Master's Thesis Researcher in Robotics
PIC4SeR
Turin, Italy
Mar 2021 - Dec 2021
Mar 2021 - Dec 2021
Turin, Italy
- •Developed an autonomous drone landing system achieving centimeter-level precision on moving Unmanned Ground Vehicles (UGVs) in real-world scenarios.
- •Designed the complete landing pipeline, integrating Ultra-Wideband (UWB) localization, perception modules, and control systems.
- •Implemented a robust EKF-based localization pipeline fusing UWB, drone, and UGV data, achieving position errors under 5 cm during chase and landing maneuvers.
- •Validated the system through Gazebo simulations and real-world field tests, proving precise and reliable landing under dynamic conditions.
Education
Master's Degree in Mechatronics Engineering
Polytechnic University of Turin
Turin, Italy
2019 - 2021
2019 - 2021
Turin, Italy
- •Final Grade: 110 with honours/110 (GPA: 4.0/4.0)
- •Won first place at Challenge@Polito: Artificial Intelligence by developing an AI NLP application that automatically generates concept maps from texts to support students with learning disabilities. The project later evolved into a startup, which raised over €1.5 million.
Bachelor's Degree in Mechanical Engineering
Polytechnic University of Turin
Turin, Italy
2016 - 2019
2016 - 2019
Turin, Italy
- •Final Grade: 110/110
Projects
Improving UMI with EKF Sensor Fusion for Embodied AI Data Collection
An improved Universal Manipulation Interface (UMI) that integrates automatic task splitting and EKF-based fusion of GoPro IMU data with ARUCO marker localization, outperforming ORB-SLAM in challenging, feature-sparse settings like agricultural fields. The video shows both data collection and the learned diffusion model’s inference.
Single and Dual-Arm Robotic System for Cucumber Harvesting
A modular, multi-arm robotic system for automated cucumber harvesting, integrating perception, task planning, and control pipelines orchestrated by Behavior Trees, achieving sub-centimeter accuracy for cucumber pose and size estimation.
Initially validated in controlled lab environments, the system was then extended to other crops and successfully deployed in real-field scenarios.
Robo-Boy
Robo-Boy is a mobile-first web app for controlling ROS2-based robots, including drones, manipulators, and ground robots.
It offers multi-platform control with customizable gamepad layouts, real-time WebSocket communication, and 3D visualization with point cloud and URDF model support.
Autonomous Drone Landing on Moving UGVs
An autonomous drone landing system achieving centimeter-level precision on moving UGVs, using an EKF-based localization pipeline integrating drone, UGV, and Ultra-Wideband (UWB) data.
Designed perception, pose estimation, control, and state machine pipelines for chase and landing, validated through Gazebo simulations and real-world field tests.
Publications
A Modular Robotic System for Single and Dual-Arm Cucumber Harvesting
Gennaro Scarati, Óscar Palacín, Mario Cavero-Vidal and Néstor García
Iberian Robotics Conference (ROBOT) - IEEE Proceedings
Accepted for presentation at ROBOT2025 to be held in November 2025 in Porto. This work presents a modular robotic system designed for automated cucumber harvesting using both single and dual-arm configurations.
Advances on Affordable Hardware Platforms for Human Demonstration Acquisition in Agricultural Applications
Alberto San-Miguel-Tello, Gennaro Scarati, Alejandro Hernández, Mario Cavero-Vidal, Aakash Maroti, and Néstor García
European Robotics Forum (ERF) - Springer Proceedings in Advanced Robotics (SPAR)
This paper presents advances on the Universal Manipulation Interface (UMI) for agricultural robotics, focusing on improving human demonstration acquisition for embodied AI models through task sample extraction and EKF-based trajectory generation for fruit harvesting applications.