Thesis and Projects

We offer several Master Thesis, Bachelor Thesis as well as projects that can be done as practica (see Lectures). Please see the list below. Please directly contact the supervisor(s) if you are interested in a specific topic.

Available Thesis Topics
Titel	Topic	Supervisor	Type	Link
Representation-Invariant Latent Spaces for Robot Video Learning	Meta-Learning	Niklas Freymuth	MSc Thesis / Research Project	Link
End-to-End Differentiable Grasping Network	Grasping and Manipulation	Ngo Anh Vien	MSc Thesis / BSc Thesis / Research Project	Link
Automatic Mesh Refinement us-ing Graph Neural Networks	Deep Geometric Meshing	Niklas Freymuth	MSc Thesis / Research Project	Link
Inverse Dynamics Learning for Pneumatic Muscular Robots	Dynamics Learning; Recurrent Neural Networks	Vaisakh Shaj, Dieter Büchler (Max Planck,Tübingen)	MSc Thesis	Link
Advanced Entropy Control for Efficient Exploration in Deep Reinforcement Learning	Reinforcement Learning	Fabian Otto, Maximillian Hüttenrauch	MSc Thesis	Link
Benchmarking Model-Based RL Algorithms	Reinforcement Learning	Jeyhoon Maskani	MSc Thesis / BSc Thesis / Research Project	Link
Model-Based RL using Recurrent Kalman Networks	Reinforcement Learning	Jeyhoon Maskani	MSc Thesis / Research Project	Link
Probabilistic Forget Gates For Changing Dynamics	Time Series Modelling	Vaisakh Shaj	Bachelor Thesis	Link

Assigned Thesis Topics
Titel	Topic	Supervisor	Student	Typ	Link
Data Augmentation for Neural Processes	Meta-Learning	Michael Volpp	Anushka Vashishtha	External Master Thesis with Robert Bosch GmbH	Link
Residual Model Learning using Action-Conditional Recurrent Kalman Networks	Dynamics Modeling	Niels van Duijkeren, Robert Krug, Vaisakh Shaj, Philipp Becker	Moritz Reuss	External Master Thesis with Robert Bosch GmbH	Link
Kalman Mixture Networks	Time Series Modeling	Chen Qui, Maja Rudolf, Philipp Becker, Vaisakh Shaj	Giao Nguyen Quynh	External Master Thesis with Robert Bosch GmbH
Improved Trust Regions for Adversarial Imitation Learning	Imitation Learning, Trust-Region Optimization	Philipp Becker, Fabian Otto, Hanna Ziesche		MSc Thesis	Link
Imitation Learning for Versatile Robot Manipulation	Imitation Learning	Philipp Becker, Onur Celik	Maximilian Li	Master Thesis
Trajectory Segmentation for Robot Manipulation Datasets	Grasping and Manipulation	Ge Li	Xin Tian	Bachelor Thesis	Link
Grasp Adaptation for Robotic Manipulation	Grasping and Manipulation	Gerhard Neumann, Vien Ngo, Hanna Ziesche	Jun Wang	MSc Thesis, Research Project	Link
Efficient Gradient-Based Variational Inference with GMMs	Variational Inference	Oleg Arenz	Zihan Ye	Master Thesis	Link
Recursive Surrogate-Modeling for Stochastic Search	Stochastic Search / Regression	Maximilian Hüttenrauch	Klaus Philipp Theyssen	Bachelorarbeit	Link
Long-Term Prediction for Pushing and Grasping with Deep Reinforcement Learning	Grasping and Manipulation	Gerhard Neumann, Hanna Ziesche, Vien Ngo	Mohammed Abdalqader	Master Thesis	Link
Learning to Assemble using 6D Pose Estimation for Insertion Tasks	Grasping and Manipulation	Gerhard Neumann, Hanna Ziesche, Vien Ngo	Artur Eckard	Master Thesis	Link
Intrinsically Motivated Deep Reinforcement Learning	Deep Reinforcement Learning	Vaisakh Shaj	Andreas Boltres	Masterarbeit	Link
Contextual Policy Search for Robot Table Tennis	Reinforcement Learning	Ge Li	Dongzhuoran Zhou	Master Thesis	Link
Learning to Pick specific Objects in a Heap	Grasping and Manipulation	Gerhard Neumann, Vien Ngo, Hanna Ziesche	Oussama Zenkri	Master Thesis	Link

Finished Thesis
Titel	Topic	Supervisor	Student	Type	Link
Actor-Critic Methods for Robotics	Reinforcement Learning	Philipp Becker, Onur Celik	Roman Freiberg	Bachelor Thesis	Link
Deep Reinforcement Learning under Partial Observability using Kalman Filtering	Reinforcement Learning	Philipp Becker	Carmen Eitel	Master Thesis	Link
Deep Reinforcement Learning of Versatile Behavior	Reinforcement Learning	Onur Celik	Dhimiter Pikuli	Bachelor Thesis	Link
Inverse Reinforcement Learning for Highly-Versatile Behavior	Inverse Reinforcement Learning	Philipp Becker	Niklas Freymuth	Master Thesis	Link