Lehrstuhl für Angewandte Softwaretechnik
Chair for Applied Software Engineering

On this page you can find a list of open topics for bachelor's and master's theses as well as topics for guided research and IDP's. 
Please also check the Research Fields and the individual pages of the members of the scientific staff.  
For your thesis, we recommend to use our Latex template. For thesis with Juan Haladjian please use: Thesis template.

 

Masterthesis
Assembly of a 9-axis IMU recording device
Advisor
Juan Haladjian
Date
26.02.2018
If you have experience on hardware design, you are welcome to apply to this project! We are looking for a device able to record acceleration, angular velocity and magnetic fields (9-axis) at high speeds on an SD card. We currently have a device able to do so based on InvenSense's ICM20602 and uses an external magnetometer. The goal of this thesis is to adapt / extend or assemble a new device using on the most modern  Invensense's MPU9250 IMU (or similar to be agreed on).
Masterthesis
Wearable Activity Recognition
Advisor
Juan Haladjian
Date
03.03.2018
My field of research is on machine learning with wearable sensors. One possible application field is sports (e.g. counting the amount of repetitions of an exercise, measuring athlete’s performance), another is veterinary medicine (activity recognition for animal: eating, walking, resting). If you have access to a specific animal or perform a specific sport and want to automatically extract context information, we could work on it together. Feel free to contact me with your idea.

For more information about applying to this thesis, visit my website.

Masterthesis
Embedded development of a sports application.
Advisor
Juan Haladjian
Date
03.03.2018
We are developing a virtual coach for goalkeeper training with a famous German goalkeeper. We have a wearable device with an integrated 9-axis IMU (accelerometer, gyroscope and magnetometer). The goal of this thesis is to develop a firmware (in plain C) that performs computations (e.g. mean, correlation, fast fourier transform) on the wearable device and sends the results over to an iOS application. Our wearable device is based on InvenSense's ICM20602 and uses an external magnetometer.

For more information about applying to this thesis, visit my website.

Bachelor / Master
Routing algorithms for multi-modal intelligent transportation systems
Advisor
Research, implementation, and evaluation of multi modal routing algorithms for intelligent transportation systems.
Bachelor / Master
Step-wise exercises with interactive help tutorials in ArTEMiS (https://artemis.ase.in.tum.de)
Advisor
Many exercises include multiple parts that depend on each other. Then it is impossible, difficult and/or demotivating for students to continue with the 2nd or 3rd part if they were not able to solve the 1st part. It is also misleading if the 1st part is finished and the student gets the feedback that e.g. 8 out of 12 test cases still fail. In this thesis, you will extend ArTEMiS so that instructors can easily add multiple parts for exercises. In addition, ArTEMiS should allow student to receive automatic help in form of live tutorials for exercise parts that they don't understand or cannot solve. While they cannot obtain the full points any more, they can work on the other parts as well and learn the system and the theory behind the concept through interactive live tutorials (comparable to an on-boarding in a mobile app). The thesis should also evaluate if these improvements help and motivate students to achieve a better learning experience. ArTEMiS is open source and available on https://github.com/ls1intum/ArTEMiS
 
Bachelor / Master
Automatic Grading of Team-based Exercises using Version Control Metrics in ArTEMiS (https://artemis.ase.in.tum.de)
Advisor
Until now, ArTEMiS only supports individual coding exercises with automatic assessment based on version control and continuous integration. This thesis should extend ArTEMiS in a way that it also supports team exercises. The first easy step is to provide access to multiple students when one of the team members starts the exercise. The second more difficult task is to find the right model to measure and grade the contribution of each single student to the completion of the exercise. In team exercises, all students should contribute and not only a single student. While there are different possibilities to measure the contribution (e.g. number of commits per student, number of passing test cases per student), such a model also influences the motivation and the behavior of the students. Therefore, it is important to find a fair assessment model that does not demotivate students and that focuses on collaboration and not competition within a team. ArTEMiS is open source and available on https://github.com/ls1intum/ArTEMiS
Master
Semi-Automatic Grading of Modeling Exercises using Machine Learning on the ArTEMiS platform (https://artemis.ase.in.tum.de)
Advisor
The objective of this thesis is to design a process to semi-automatically grade modeling exercises during a lecture using machine learning algorithms. The first part is to integrate a simple and easy to use UML editor into the ArTEMiS platform and allow teaching assistants to grade correct and wrong parts directly in the browser. The machine learning algorithm should then used the inputs of the teaching assistants and automatically apply it on all other student solutions so that only a few corrections are needed to grade all solutions. This idea should be implemented and then evaluated in a real course. ArTEMiS is open source and available on https://github.com/ls1intum/ArTEMiS