Physics

Title of thesis: SD Map Localization: A Deep Learning Approach: Using deep learning models to accurately localize vehicles on Standard Definition maps using GNSS and INS data
Abstract: Accurate localization is critical for the safe and efficient operation of autonomous vehicles, enabling precise navigation and real-time decision-making. This thesis focuses on improving Standard Definition (SD) map-based localization, by leveraging deep learning techniques. The research addresses key questions, including how to optimally encode SD map data and sensor data, particularly the GNSS and INS sensor, for deep learning, and train models to perform accurate map matching and vehicle localization along the correct road segment.
Examiner: Mats Granath Supervisors: &nbsp;Axel Beauvisage and Junsheng Fu&nbsp; Opponents: Artur Gasparyan and Ruiqi Qiu&nbsp;
Location of presentation: <a rel="noopener" href="https://maps.chalmers.se/#495a8bdb-ae0d-4c94-8be5-68240f9a5686" target="_blank" title="Chalmers maps" data-anchor="#495a8bdb-ae0d-4c94-8be5-68240f9a5686">von Bahr, Soliden floor 3</a>
Online: <a rel="noopener" href="https://chalmers.zoom.us/j/7419214619?omn=61865079465" target="_blank" title="Zoom" data-anchor="?omn=61865079465">Zoom</a>

Title: Retrieval-Augmented Generation with Regional Medical Guideline
 Abstract: As of today, medical knowledge from clinical and experimental studies are compiled and accessible through medical guidelines. They are represented as unstructured free text, making navigation and retrieving relevant information a challenge. In recent years, Large Language Models (LLMs) powered by advanced artificial intelligence (AI) and machine learning (ML) technologies, have shown promise in addressing some of healthcares current challenges, such as time-consuming working tasks. Previous studies has shown great potential with the combination of LLMs, vector databases and the Retrieval-Augmented Generation (RAG) technique. By supplying LLMs with contextual information, it can enhance the accuracy and improve the generated output. LLMs have the capacity to both comprehend and generate human-like text and can serve as a key component of the foundation for a digital assistant with access to the medical guidelines. This master’s thesis explore alternative ways to store, retrieve and present information from Region Västra Götalands (VGRs) medical guidelines. The information presented in the guidelines is also evaluated and tested as a data source to answer questions from medical exams. The working process included Natural Language Processing (NLP) techniques, such as chunking and vector embedding, to extract and embed text from VGRs medical guidelines into a vector database. The database was queried for relevant information to use as context for different LLMs to answer multiple-choice questions from medical exams. Different chunk sizes, embedding functions and LLMs were compared to try and optimize the results. Little to no improvement is shown when using VGRs medical guidelines as contextual information for LLMs to answer medical question. At most, the retrieved data were useful to answer only 28% of the medical questions. The results suggests that VGRs medical guideline might not be sufficient to answer complex medical questions, such as those found on a medical exam. This thesis wants to give a first insight in the use of VGRs medical guidelines as a data source for AI purposes. It shows the need for further exploration and more ways of evaluation.
 Examiner: Mats Granath Supervisor: Magnus Kjellberg&nbsp; Opponent(s): TBD
Location of presentation: <a rel="noopener" href="https://maps.chalmers.se/#495a8bdb-ae0d-4c94-8be5-68240f9a5686" target="_blank" title="Chalmers maps" data-anchor="#495a8bdb-ae0d-4c94-8be5-68240f9a5686">von Bahr, Soliden floor 3</a>
Online: <a rel="noopener" href="https://chalmers.zoom.us/j/7419214619?omn=61865079465" target="_blank" title="Zoom" data-anchor="?omn=61865079465">Zoom</a>

Architecture and Civil Engineering

Bachelor’s thesis Chemistry and Chemical Engineering

Chemistry and Chemical Engineering

Communication and Learning in Science

Computer Science and Engineering

Electrical Engineering

Industrial and Materials Science

Life Sciences

Mathematical Sciences

Mechanics and Maritime Sciences

Microtechnology and Nanoscience

Space, Earth and Environment

Technology Management and Economics

Presentation of master thesis

Choose subject

Course syllabus for

Subject

Each academic year contains different courses

See programme page at universityadmissions.se

Educational area

Permissible substitutions of compulsory courses

To see your programme syllabus – select the academic year in which you started the programme.

[en: page.studyprogrammepage.syllabusesview.coursesdescription]

Presentation of master thesis

Master thesis presentations at Physics

September 20 2024

10:00- 11:00 Sameer Jathavedan & Sheik Meeran Rasheed Abdul Rahuman, MPCAS

11:00- 12:00 Albin Persson & Jonathan Pettersson, MPCAS