I finished my PhD on multimodal perception of humans for mobile robots in 2020 at the Social Robotics Lab, University of Freiburg, Germany. My research was funded by the EU FP7 project SPENCER, which involved deploying a first-of-its-kind autonomous service robot in a crowded airport environment. After that, I worked on two publicly funded projects in the intralogistics sector, involving heterogeneous fleets of mobile robots in ILIAD (where I lead a work package on human-aware AGV fleets), and a mobile manipulator in DARKO (where I led the 3D scene understanding work package). Since 2023, I have also been leading a large-scale academic research collaboration (5 PhD students) on context understanding for autonomous systems. Most recently, in summer 2025, I joined a team working on an end-to-end pipeline for urban automated driving.
Earlier on in my career, over a duration of 5 years in my free time, I developed in C++ a 2D real-time strategy game called Sorades – Die Befreiung, which won the Newcomer Award at the German Game Developer Award 2006. My further interests include photography, travelling, and aviation.
News
September 2025: Together with PhD student Kadir Yilmaz and master student Mert Arslanoglu from Prof. Leibe’s computer vision group at RWTH Aachen we got a paper accepted at CoRL 2025! In this collaborative effort, we introduce a novel benchmark for 3D object referral via pointing gestures. We also introduce a novel masked transformer-based approach for 3D instance segmentation of previously unseen objects via monocular camera-based pointing gestures, where the target object does not require to be visible in the current camera view. This key technology can greatly enhance human-robot interaction in open-world settings.
August 2025: My PhD student Stephanie Käs at RWTH Aachen, together with her students, got a paper accepted at IEEE RO-MAN 2025! In this work, we explore how well 2D vision foundation models (VFMs) and vision-language models (VLMs) cope against a classical skeletal-based approach on the task of full-body gesture recognition. We also introduced a challenging new benchmark dataset comprising 15 different gesture classes for mobile manipulation tasks.
July 2025: After over 12 years of advancing research on mobile robotics for indoor environments, with the end of DARKO, I will soon be switching teams and extend my horizon by working on an end-to-end autonomous driving stack for urban automated driving. Looking forward to the new challenge!
July 2025: Final project demonstration of our EU H2020 project DARKO at the Automatica 2025 Fair in Munich! As part of the AI.Society area, our exhibition was very close to the large Neura Robotics booth that was in the center attention of many visitors of the fair. Onboard our DARKO mobile manipulator robot, we showcased advanced 3D human and object perception capabilities using fisheye cameras and 3D lidar running in real-time, along with a stunning demo of an elastic manipulator throwing objects over 5-7 meters distance at high accuracy.
June 2025: The PhD students I am co-supervising together with Prof. Bastian Leibe at RWTH Aachen got two CVPR 2025 workshop papers accepted! One full paper on open-vocabulary 3D instance segmentation using Gaussian Splatting (CVPR 2025 OpenSUN3D workshop), and an extended abstract on leveraging 2D foundation models for 3D segmentation (CVPR 2025 Transformers for Vision workshop).
June 2025: Final stakeholder and review meeting of our EU H2020 project DARKO at the KI.Fabrik at Deutsches Museum in München! In the last few weeks, we have been busy deploying and integrating our latest AI-based methods on the DARKO robot platform and their embedded Nvidia Jetson accelerators. For me, tuning the real-time performance of our perception algorithms was a great chance to dig deeper into the ONNX & TensorRT frameworks. Check out the picture gallery on the DARKO website for some impressions of our integration week!
May 2025: Busy review season this year! I have been reviewing papers for ICCV 2025, the CVPR 2025 Multimodal Learning & Applications Workshop, and serving as associate editor for the computer vision & datasets topic at IROS 2025.
February 2025: Together with our collaborators from the EU H2020 project DARKO, we will be presenting an accepted paper at the European Robotics Forum 2025 on hierarchical risk and context-aware robot navigation in complex intralogistic environments.!
January 2025: Got two co-authored papers accepted at ICRA 2025 (on monocular fisheye-based 3D human pose estimation, and on joint human full-body and trajectory prediction). The first is a collaboration with my PhD student Stephanie Käs from RWTH Aachen University and further colleagues, the second one a joint work with our former master student Nisarga Nivaladi from Uni Freiburg.
September 2024: Together with our visiting PhD student Elisa Stefanini from Università di Pisa and collaborators from Bosch Research, we got an article accepted at IEEE Robotics and Automation Letters (RA-L) on efficient context-aware model predictive control for human-aware navigation of mobile robots!
June 2024: Our co-supervised PhD student Jens Piekenbrinck from Prof. Leibe’s computer vision group at RWTH Aachen got a paper accepted at the CVPR 2025 Multimodal Learning and Applications (MULA) workshop with the title “RGB-D Cube R-CNN: 3D Object Detection with Selective Modality Dropout”!
January 2024: We recently kicked off a large-scale academic collaboration on 3D context understanding for autonomous systems. As collaboration lead, I am responsible for a team of 5 PhD students on advanced 3D scene understanding tasks. Have a look at this news article for more details!
October 2022: Check out the latest episode of the Bosch Global Podcast “From Know-How to Wow” on our research project DARKO!
June 2022: Our Semantic SLAM team at Bosch ranked first in the public leaderboard of Semantic SLAM task in the passive actuation category at the Robotic Vision Scene Understanding Challenge held at the Embodied AI Workshop at CVPR 2022! In our semantic SLAM team, which I joined part-time in 2021 and 2022, I am contributing to 3D object recognition and data association topics and the 3D data labelling pipeline.
January 2022: Our publicly funded EU H2020 project DARKO is featured in a story on bosch.com: “DARKO – Object-throwing robots with people skills”.
August 2021: Check out our IROS 2021 paper “Cross-Modal Analysis of Human Detection for Robotics: An Industrial Case Study” on arXiv!
July 2021: Successful conclusion of the EU H2020 project ILIAD which I was involved in as a work package leader for Human-Aware AGV Fleets!
January 2021: Kick-off of the EU H2020 project DARKO, where I am leading WP2: 3D Perception and Scene Understanding.
April 2020: Successfully defended my PhD thesis on multi-modal human detection, tracking and analysis for robots in crowded environments at the University of Freiburg!
June 2019: Got a paper accepted at the 3D Scene Generation workshop at CVPR 2019 in Long Beach, CA! I will also be attending RSS 2019 in Freiburg.
May 2019: I have been awarded an Outstanding Reviewer Award at ICRA 2019 in Montreal, Canada!
September 2018: Two workshop papers (joint work led by István Sárandi, RWTH Aachen) accepted at ECCV’18 and IROS’18 on Synthetic Occlusion Augmentation for 3D Human Pose Estimation. We won the PoseTrack 3D challenge at ECCV!
April 2018: Joint paper with RWTH Aachen on Deep Person Detection in 2D Range Data accepted for Robotics and Automation Letters!
January 2017: Kick-off of EU H2020 research project ILIAD on intralogistics, where I am work package leader for Human-Aware AGV Fleets.
September 2016: My PhD advisor Kai, our robot SPENCER and I were featured as invited guests in an episode of the Tigerentenclub, a popular German TV show targeted at school children. SPENCER performed well in the spotlight in front of five cameras!
June 2016: I joined Bosch Corporate Research in Renningen, Germany, as a full-time researcher in robotics and perception!
May 2016: Last month, the SPENCER project ended with a very successful demonstration at Amsterdam-Schiphol airport. The SPENCER robot is now resting peacefully in its wooden storage box. A TV report on our project, also featuring me, was recently shown on German television. Goodbye SPENCER!
February 2016: The end of the SPENCER project is approaching quickly and we are preparing for next month’s final demonstration at Amsterdam-Schiphol!
December 2015: First week-long deployment of the SPENCER robot at Schiphol for initial testing (pictures!). People were very curious and overall reacted very positively. Also lots of press coverage, e.g. by Heise Online, The Telegraph, Daily Mail or phys.org.
November 2015: SPENCER robot is coming to Freiburg, where my colleagues Luigi, Billy and I will do the final preparations for the first Schiphol deployment. My main focus will be on making the robot operate safely among humans in crowded spaces.
October 2015: I presented my paper on full-body human attribute recognition in RGB-D in one of the computer vision sessions at IROS’15 in Hamburg. Later on, headed to Toulouse for SPENCER integration week IV.
April 2015: Just came back from an eventful SPENCER Integration Week III (see gallery). Also check out our new video on Real-Time Multi-Modal People Tracking in Crowded Environments! The code is available on GitHub.
January 2015: My paper on Real-Time Full-Body Human Gender Recognition in (RGB)-D Data has been accepted for presentation at ICRA’15. Looking forward to visiting Seattle in May! I also submitted a follow-up paper to IROS 2015 in Hamburg, Germany.
October 2014: After submitting a paper to ICRA’15, it is now time to prepare for the SPENCER integration event in Toulouse! I’ll be heading there twice, during the first meeting we will set up the basic robot software infrastructure in a smaller team. Looking forward to working with the robot for the first time! Also going to Aachen for people tracker integration. Update: Demo videos of single-person and group guidance now online!