Thesis and Internships
Here you can find a list of available Bachelor and Master thesis available, along with internships. There may be also novel projects available, so please check with me if you are interested in the topics I work on.
General rules for taking an internship or thesis work with me
I am always happy to work with Bachelor and Master students, and I am glad that you are considering one of the topic I propose. Undergoing a thesis work is a key step in your education, and it is a challenging yet rewarding experience. The following rules are meant to be a guideline so that both you and me can carry the project in an efficient manner.
- Take your time: a good bachelor thesis requires around 3 months, while a Master thesis can take 6 months or even more. If you are interested in one or more topics, ask in advance, even if you have not finished your exams yet. Moreover, if you are interested in one of the topics listed in this pages, say it upfront.
- Be efficient with communication: when reporting your work, one email per month is too few, one per day is too much. Be efficient in what you communicate and how you do so.
- I am not gdb (or similar): do not send me code saying “it does not work”. Explain what you have tried, on what circumstances you experience any problem. Be proactive, be clear.
- How to write: we will use LaTeX, which may seem difficult at the beginning but you will benefit a lot from it at the later stages. A good tool to share the thesis work is Overleaf.
- The Internship/thesis is YOUR responsibility: I will not remind you about deadlines, documents to fill, bureaucracy and alike. It is your responsibility to perform these tasks.
Internships with Industries
| Title | Description | Partners |
|---|---|---|
| Distributed Liquid Computing Dashboard | Design and develop a dashboard for Distributed Liquid Computing to monitor the status of a distributed liquid computing system, gathering data from devices and visualizing them. The dashboard also supports the definition of new pipelines. | |
| Self Supervised Learning | Most data obtained from wearable devices is unlabeled. In this thesis, we want to explore how it is possible to label unlabeled data by utilizing data obtained from different sensors, and complement that of others on the same human body through Self Supervised Learning. | |
| Homeassistant data collector | Obtain data from homeassistant. Collect status of sensors and save them so that it can be later possible to perform human-activity-recognition tasks. | |
| Data analysis on user behavior | Through the analysis of user reports in a deployed crowdsensing application, this thesis focuses on understanding patterns about user communication to the central platform. | Comunichiamo (BO) |
| Orchestration Algorithm | Design, implement and evaluate an orchestration algorithm for distributed Liquid Computing | |
| Metadata definition for Fluid Computing | Study and design the metadata for tasks and devices' capabilities in distributed Liquid Computing | |
| TinyML offloading Networking optimization | Starting from the code available at https://github.com/UBICO/SCIoT, develop and test new communication protocols for the system | |
| ESP32 With Edge Impulse | This thesis should test the limits of on the thesis Edge Impulse specifically for on the thesis ESP32 devices.- network complexity- model size- comparison with custom models | |
| Bicycles determine the road condition | Determine the road condition based on users which perform specific actions in certain locations. Start from the thesis done at UNIBO. | UNIBO |
| application | Need an omnicomprehensive application to collect data from smartphones. This ideally should go into the backend developed as part of on the thesis Crowdsensing - backend and frontend. Need to check with comitato etico beforehand. | UNIBO |
| economical benefits | It is also needed to assess the economical benefits of crowdsensing. Simple analysis which takes into account costs for infrastracture based solutions and for crowdsensing applications. | UNIBO |
| Packet fall detection | The goal of this internship is to develop a vision system to track packets and boxes which fall from pallets, to notify operators to fix the issue. | Kosme Srl |
| Comunichiamo text | Improve the regressor based on the GoodIT 2023 paper. | Comunichiamo SrL |
| Business Intelligence System Development | Digitalization and Integration of the ERP, MES and BMS systems. This allows the use of advanced data analysis techniques on the data obtained. | Ingranaggi Moreali (RE) |
| Text and Drive | Use the rear camera of the smartphone to detect steering wheel, Gear and other objects.Check differences between night and day modelsMeasurements on the inference time, Model size on the thesis Activity recognition - benchmarkPersonalize models with federated learningMake a short video which shows each frame being classified and what does this mean for overall classification. We can also try with different models such as yolo or smartdetGet a dataset while actually texting and driving but only with images taken as the passenger, then flip them horizontally to mimic the driver seat | UNIBO |
| Packets on pallet stability | The goal of this internship is to analyze a vast dataset of issues related to pallets building, to find patterns about issues related to packets and boxes falling from them. | Kosme SrL |
| Reconfiguration Algorithm | Starting from a pipeline orchestration in Liquid Computing, design, implement and evaluate a reconfiguration algorithm leveraging digital twins for simulations | |
| Task decomposition | Starting from monolithic code, automatically decompose it into multiple, simpler tasks that can be computed in distributed nodes. | |
| Scheduler and Optimizer Simulator | This internship aims to develop a scheduler and optimizer for the association of workers to specific machines in an industrial context. | Ingranaggi Moreali (RE) |
| Packets on pallets stability | The goal of this internship is to analyze a vast dataset of issues related to pallets building, to find patterns about issues related to packets and boxes falling from them. | Kosme Srl |
| benchmark | Provide a framework to test different activity recognition algorithms.Work: research already published algorithms, develop a layer to homogenize data, and test all developed algorithms. | |
| Data digitalization | This internship aims to digitalize all the data available on the machines of the industry. Currently, not all of it is digital, as some data is taken manually. Through the use of sensors and system integration, the data will be brought to a common shared platform. | Ingranaggi Moreali (RE) |
| TinyML Split Computing library | Starting from https://github.com/UBICO/SCIoT, develop an edge server capable of handling the split computing requests. Publication available at https://doi.org/10.1109/CCNC51664.2024.10454775 | |
| different devices | Test how different devices perceive the same data. Here the objective is to develop a simple application which collects data and records it, and compare it against different devices.The thesis should also analyse the state of the art to find which data is sensed by crowdsensing applications and categorize it, so that it is possible to compare devices against useful data. |