Each student has to choose one of the offered topics from the area of deep learning / robot learning / deep reinforcement learning / deep imitation learning.

The students need to implement one or several algorithms and evaluate them against available baselines on standard benchmark tasks as well as on (custom-made) physically realistic simulations and/or a real robot platform.

The experiments have to be documented in a report.  

Students will work in teams of 2.

It is recommended to take this course together with the seminar “Deep learning for robotics” where the students will acquire the required background on the literature.

Arbeitsaufwand: 180h

Präsenzzeit: 15h

Projektarbeit: 135h

Report schreiben + Präsentation vorbereiten: 30h

Ein Rücktritt ist innerhalb von zwei Wochen nach Vergabe des Themas möglich.

Weitere  Infos in ILIAS