Filippo Samorì

Project realized at Eurecat during my robotics research internship. It involved developing a Projected Augmented Reality (PAR) framework to display real-time information in the robot's workspace, such as end-effector trajectories, highlighted objects, and task instructions. ​Additionally, I explored the MoveIt Hybrid Planning framework, developing custom plugins that enabled a seamless transition from simulation to the physical robot, resolving several limitations present in the official plugins.

Implemented a colorization algorithm using TensorFlow. Designed and trained a Deep Neural Network based on a U-Net architecture to colorize grayscale images.

Implemented a SLAM algorithm on a Crazyflie drone using ORB_SLAM3. The architecture consisted of a server running ROS 2 with ORB_SLAM3 and the drone transmitting data via Bluetooth and Wi-Fi, achieving real-time performance.

Universal Robot 5 modelling and simulation using 20-sim. The model was developed using a modular architecture based on bond graphs. Implemented a control law to operate the robot and validate simulation behavior.

A project focused on detecting and localizing humans during robotic exploration. Used YOLOv11 for detection and an Extended Kalman Filter for localization.

Simulation of the Tiago robot in Gazebo. The project involved using the ready to use simulation environment provided by the unversity and implementing a simple task using ROS 2. The robot was able to navigate in the environment using the ROS 2 Navigation Stack and to perform a pick and place of cubes detected using fiducial markers. The motion planning was implemented using the MoveIt Framework.

Project realized during the optimal control course at University of Bologna. The project involved developing the optimization algorithm for planning trajectories for a 2 dof robotic arm (double pendulum) with no actuation in the second joint. Newton's method and LQR has been developed inside the project.

Project realized during a 24h hackaton together with Natalie Morandi. The idea was to create a smart water meter integrating IoT technology to monitor water usage and provide real-time data to the user, enabling also predictive maintenance for the water supply system.

Project realized during the computer vision course at University of Bologna. The project involved developing an algorithm for barcode detection using standard computer vision approaches. The dataset was a black and white set of images.

Small project realized for fun. The game is a simple implementation of the classic Snake game, playable in the terminal. It was developed in C and uses only terminal features for the GUI.

Project realized by a team of six during my fourth year of high school. Developed in collaboration with Fores Engineering, the project involved creating a sun-tracking solar panel. Together with my classmates from the electronics course, we implemented the logic and the circuitry. We used Arduino and designed a custom expansion board to connect the sensors. We won first place.

Project realized with my classmate Federico Cobianchi during our fourth year of high school. The project involved the development of a simple automatic greenhouse for a small plant. We used Arduino with various sensors to detect when the plant needed water.