OPEN POSITIONS - MASTER THESIS

Application deadline: 31/01/2022

Deep Neural Network for Map Interpretation and Segmentation

Develop an algorithm able to recognize and segment rooms composing an indoor environment map obtained from SLAM algorithms and Lidar sensors, exploiting computer vision and deep neural networks.

Welcome Skills

Deep Learning, Computer Vision, Python, ROS2

Supervisor:
M. Chiaberge, A.Eirale

Code: 

PIC4SeR-01-2022

Visual odometry algorithm in light denied environments

Phase 1: Analyse the behaviour of already exixting visual odometry algorithms with infrared images as input. Phase 2: Considering the results of Phase 1, select an algorithm and try to optimize it for the navigation in environments with no light.

Welcome Skills

ROS/ROS2, Gazebo, Python, Computer Vision

Supervisor:
M. Chiaberge, C. Boretti

Code: 

PIC4SeR-02-2022

Realistic Depth Images in Simulation: learning noise of depth cameras

Generate realistic depth images in simulated environments by learning the noise existing in real camera sensors. The study enables realistic training and testing conditions for any robotic task leveraging depth images. It involves the use of simulation engines (Gazebo) as well as the study of computer vision techniques and metrics and generative deep learning methods. The student will become confident with typical camera sensors for robotic applications (Intel RealSense, ZED2).

Welcome Skills

ROS/Gazebo, Python, Computer Vision

Supervisor:
M. Chiaberge, F. Salvetti

Code: 

PIC4SeR-03-2022

UWB anchors deployment system for space applications

Design a compact and effective system to deploy UWB anchors. This system should be mounted on a UGV that will drop anchors on the ground on user’s request. Also, the device should be integrated in a ROS2 environment for control and communication with other components of the system.

Welcome Skills

Mechanical Design, ROS2, Python/C++

Supervisor:
M. Chiaberge, M. Ambrosio

Code: 

PIC4SeR-05-2022

Robust autonomous landing on a high-speed platform

Optimize an existing algorithm performing autonomous landing on a moving ground vehicle using UWB technology. The improvement, using data from other sensors and/or a camera, should make it reliable and working also at higher speeds, as well as increasing its precision and performance.

Welcome Skills

ROS2, Python/C++

Supervisor:
M. Chiaberge, M. Celada

Code: 

PIC4SeR-06-2022

250g drone sensorization for indoor autonomous missions

The aim of the thesis is to extend the onboard sensors of PICCOLO (PIC4SeR’s custom 250g drone). The minimum set of sensors will include an UWB transceiver and several ToF modules needed to perform indoor autonomous missions, obstacle avoidance and mapping. The candidate will be in charge of choosing the right set of sensors, designing the interface circuit needed to communicate with the onboard computer (RPI zero) and writing a ROS2 driver to make the data available to the flight controller.

Welcome Skills

Electronics, ROS2, C++

Supervisor:
M. Chiaberge, G. Fantin, M. Celada

Code: 

PIC4SeR-07-2022

Learning Odometric Error in Mobile Robots with Neural Networks

A correct odometry signal is the foundation of any reliable autonomous navigation system. As basic goal, the study aims at collecting a numerical odometry dataset with a mobile robot, in order to learn the odometric error with a Neural Network and a supervised learning method.

Welcome Skills

ROS2, Deep Learning, Python

Supervisor:
M. Chiaberge, M. Martini

Code: 

PIC4SeR-08-2022

Hand Gesture Recognition for Home Robotics

Develop a Deep Learning based Hand Gesture Recognition system for robot assistants in a domestic environment, using a simple RGB camera. The focus is on a small set of useful gestures, in order to obtain a robust and accurate solution embeddable on devices with small computational resources.

Welcome Skills

Deep Learning, Computer Vision, Python

Supervisor:
M. Chiaberge, S. Angarano

Code: 

PIC4SeR-09-2022

Underground Tunnel Inspection

Design a payload made of the minimum amount of sensors and algorithms able to deliver autonomous navigation capabilities to UAV/UGV in challenging and unsupervised environments such as tunnels.

Welcome Skills

Python, C++, ROS/ROS2

Supervisor:
M. Chiaberge, S. Cerrato

Code: 

PIC4SeR-11-2022

Modular device for obstacle avoidance bus network

Design and developement of a bord with different obstacle avoidance sensors with a modular approach based on CAN bus communication. Particular attention must be paid to the sensors selection and their interaction.

Welcome Skills

Electronics, C, MPLAB CAN, KiCad

Supervisor:
M. Chiaberge, G. Dara, D. Gandini

Code: 

PIC4SeR-12-2022