Profile Picture

Malte Mosbach

I am a PhD candidate at University of Bonn advised by Sven Behnke. My research aims to make reinforcement learning practical for complex robot control by leveraging foundation models, scaling training, exploring student-teacher learning, developing robust representations, and more.

E-Mail me Connect on LinkedIn

Selected Publications

A complete, up-to-date list of publications can be found on Google Scholar.

SOLD: Reinforcement Learning with Slot Object-Centric Latent Dynamics

Malte Mosbach*, Jan Niklas Ewertz*, Angel Villar-Corrales, Sven Behnke
arXiv, 2024
Read the PaperBibTeX

Promptable Visual Object Manipulation in Cluttered Environments by Memory-augmented Student-Teacher Learning

Malte Mosbach, Sven Behnke
Under review
Read the Paper

Grasp Anything: Combining Teacher-Augmented Policy Gradient Learning with Instance Segmentation to Grasp Arbitrary Objects

Malte Mosbach, Sven Behnke
ICRA, 2024
Read the PaperBibTeX

Learning Generalizable Tool Use with Non-rigid Grasp-pose Registration

Malte Mosbach, Sven Behnke
CASE, 2023 (Best Conference Paper Award)
Read the PaperBibTeX

Efficient Representations of Object Geometry for Reinforcement Learning of Interactive Grasping Policies

Malte Mosbach, Sven Behnke
IRC, 2022
Read the PaperBibTeX

Accelerating Interactive Human-like Manipulation Learning with GPU-based Simulation and High-quality Demonstrations

Malte Mosbach, Kara Moraw, Sven Behnke
Humanoids, 2022
Read the PaperBibTeX Code

Transfer of Hierarchical Reinforcement Learning Structures for Robotic Manipulation Tasks

Christian Scheiderer, Malte Mosbach, Andrés Posada-Moreno, Tobias Meisen
CSCI, 2020
Read the PaperBibTeX Code