Current Master-Thesis Projects


Studio e progetto di generatore di vibrazioni in macchina automatica contatrice

Study and development of a vibrations generator in an automatic counter machine

The thesis aims to design a quick fixing system of parts in an automatic machine using shape memory alloys.

Supervisors: Prof. M. Carricato, R. Di Leva, F. Sassi
In collaboration with:
IMA Confectionery

Afferraggio e manipolazione di un oggetto virtuale generico tramite tecniche di apprendimento per rinforzo

Grasping and handling of a virtual generic object through reinforcement learning techniques

Source: DeepSym: Deep Symbol Generation and Rule Learning from Unsupervised Continuous Robot Interaction for Planning – Scientific Figure on ResearchGate. Available from:

The project involves the study and development of a software tool capable to train a robot to grasp a generic object in a virtual environment. The idea is to exploit reinforcement learning techniques in order to keep the purpose of such tool as generic as possible, capable of virtually training the robot to grasp different shaped objects with no additional costs. The work will start studying available documentation, first attempt at objects, sensors and robot virtualization and finally the definition of a reward function able to point the system to convergence. 

Supervisors: Prof. M. Carricato, S. Comari, D. Baraccani
In collaboration with:
IMA R&I S.p.A.

Sviluppo di un robot parallelo a cavi per operazioni su facciata continua

Development of a Cable-Driven Parallel Robot for operations on continuum facades

The first aim of this project is to conduct a feasibility study about the possibility of executing operations on continuum facades with a Cable-Driven Parallel Robot. Then, the mechatronic design of the manipulator will be developed by taking into account the rapid on-site installation of the machine and its task. Finally, the prototype will be assembled, and a demonstration of an operation (e.g. painting) on a wall will be conducted.

Supervisors: Ph.D. E. Idà, Prof. M. Carricato, F. Zaccaria, M. Angelini
In collaboration with:

Wire loop game challenge through reinforcement learning and force control

This project aims to solve the “Wire and loop” with a robotic manipulator provided with a force/torque sensor at the end-effector and trained by reinforcement learning. In this game, the objective is to reach the end of a 2D wire path by moving a ring fixed to the robot’s hand around it. Thanks to the intuitive idea of trial-and-error and the use of the force sensor that can perceive the ring’s contact with the wire, the robot should progressively learn how to move forward by suitably tilting the ring and by translation. The work is based on ROS and open-source libraries and proves how an industrial robot can benefit from high-level programming outsourced to an external PC.

Supervisors: Prof. M. Carricato, S.Comari, M. Angelini
In collaboration with: