Reinforcement Learning for Autonomous Driving
Learning-based control for driving tasks such as car-following, merging and highway weaving, including recurrent and multi-agent formulations.
Autonomous systems research
Paul Auerbach
Researcher in Autonomous Driving, Robotics & Reinforcement Learning
Research
What I work on
My research is centered on autonomous driving and combines machine learning, simulation, sensing and physical vehicle platforms. More broadly, I am interested in robotics and autonomous systems, especially methods that connect algorithm development with evaluation on real hardware.
Radar Perception & Simulation
FMCW radar simulation, signal processing and learning directly from radar observations, with emphasis on realistic sensing effects and interference.
Robotics & Sim-to-Real Systems
Transfer and evaluation of autonomous-driving methods on physical vehicle platforms, together with broader robotics work involving ROS 2, embedded computing, sensing, testbed development and hardware/software integration.
Publications
Selected work
Publications from 2022 onward. For citation counts, earlier work and the most up-to-date list, see my Google Scholar profile.
“A Watering Can That Never Runs Empty”: A Case Study on Human-Robot Collaboration for Professional Gardeners
V-TRAD: A Visually-Augmented Testbed for Research in Autonomous Driving
Autonomous Driving Small-Scale Cars: A Survey of Recent Development
Reinforcement Learning for Weaving Maneuvers
A Control Pipeline for Robust Lane Keeping in Model Cars
ROS2-based Small-Scale Development Platform for CCAM Research Demonstrators
About
Autonomous-driving researcher and robotics engineer
I am a research associate at the Barkhausen Institut in Dresden and a PhD Student at TU Dresden. My primary research focus is autonomous driving, particularly reinforcement learning, radar-based perception and the transfer of learning-based methods from simulation to physical vehicle platforms. Alongside this, I have a broader interest in robotics and autonomous systems.
A recurring theme in my work is the full path from an algorithmic idea to a working physical system: building simulation environments, training and evaluating learning-based policies, processing sensor data, integrating ROS 2 and embedded hardware, and validating the resulting systems on real vehicles and robotic platforms.
Beyond my autonomous-driving research, I enjoy robotics as an engineering discipline in its own right. I am particularly interested in mobile robotic systems, embedded and performance-oriented programming, Linux-based robotics software, sensing and practical hardware/software integration.
Get in touch
Interested in autonomous driving, reinforcement learning, radar sensing, robotics or autonomous systems? You can find my publications on Google Scholar or get in touch through LinkedIn.