BSc Theses in the Jacobs Robotics Group

General Information

In general, it is a very good idea to do a BSc thesis in an area where you have some experience in, which you like, and where you performed well in a related course. So, based on how things went in your 2nd year choice modules, you should select the area and also the supervisor for your thesis.

For topics in robotics, this means that you should have taken the AI and/or the Robotics course of Prof. A. Birk and that you should have performed well in them. Also, having taken a Machine Learning (ML) or Computer Vision (CV) related course and performed well in it is an option for a some of the topics from the robotics group.

As mentioned, it is important to choose your topic and your supervisor according to your 2nd year modules that you have taken that went well. A BSc project/thesis is not a good option to learn a completely new topic (e.g., trying to learn something about Deep Learning if you have no previous knowledge about it); but you should build upon existing knowledge and apply it to something new. 

To do a project/thesis in the Robotics Group, you must have

·        Passed “Intro to RIS” with a good to excellent grade

·        Passed “C/C++” with a good to excellent grade

·        Passed “Robotics” with a good to excellent grade (or as CS 3rd year student, you should at least currently take it) 
   

Topics

This is the list of topics and students for Fall 2022/Spring 2023 (seminar/thesis):

·         Detecting Cracks in Underwater Images [Flavia Tasellari]

·         Artificial Underwater Image Streams [Sindi Tasellari]

·         Underwater Stereo Processing of Divers with GAnet [Xhersila Olldashi]

·         Underwater Stereo Processing of Divers with MobileStereoNet [Sultan Sadiq Husain Siddiqui]

·         Underwater Stereo Processing of Divers with PSMnet [TAKEN by Ayush Poudel]

·         Detection of Humans in Visual and Thermal Images with Classic Computer Vision [Ilayda Güner]

·         Detection of Humans in Visual and Thermal Images with Deep Learning (A: Yolo v3) [Utkrist Karky]

·         Detection of Humans in Visual and Thermal Images with Deep Learning (B: Yolo v7) [Manish Thapa]

 

How to do a Project/Thesis

Some background information on how to do a thesis, respectively how to play the “game of science” as a BSc student can be found in the following document: The Game of Science

You should also check the CS webpage with information on the thesis and the overall process including the deadlines: https://cnds.jacobs-university.de/courses/bsc-2022/

For the time plan and the scheduling of your work, following aspects are important

·         A proper literature survey, reading, and describing the state of the art is a very crucial part of the project/thesis (see also “game of science” above).

·         For the implementation, this is much more than the coding itself (typically <50% of the implementation effort). A core part (typically >50% of the implementation effort) is doing proper experiments where you test the effects of parameter settings and where you get substantiated results.

·         The documentation of the implementation as well as of the results, i.e., the proper write-up with nice illustrations, tables, figures, etc., is also a major effort, which should not be underestimated.

It makes a lot of sense to not just start with the implementation, but to do a proper literature survey and to write up the state of the art first, while starting with the implementation. Also, documenting the implementation and (step by step) results in a continuous manner is strongly recommended.

  

Research-Oriented Thesis

For excellent students who are interested in getting a first dive into research, it is an option to consider the following topics. For these topics, there is a very high risk of failure!!! Please note that this only make sense if (a) you understand the topic presented in the slides, (b) you are willing to work yourself into the topic and to read some background material, (c) you have excellent theoretical skills, and (d) you are willing and capable to work independently on a challenging topic. As a reward, there is a high likelihood that a scientific publication is the outcome.

·         RESEARCH TOPIC: B-Scheduling of behavior processes

·         RESEARCH TOPIC: Underwater Image Enhancement with Sea-Thru

·         RESEARCH TOPIC: Machine Recognition of “Interesting” Underwater Video Sequences [Artem Karamyshev]

·         RESEARCH TOPIC: Photogrammetry on Dissimilar Images