Research
Research at the Chair of Robotics explores the nexus of physical artificial intelligence, network softwarization, and machine learning. We focus on developing adaptive control strategies, deterministic algorithms, and low-latency communication frameworks to enable seamless real-time interaction between humans and robots. Grounded in the principle that true innovation requires the interplay of theory, software, and hardware, our algorithms are rigorously validated on advanced experimental testbeds. Through close collaboration with academic and industrial partners, we translate cutting-edge research into resilient key technologies for future automation and collaborative robotic environments.
Research focuses
At the Chair of Robotics, our research advances intelligent, networked, and physically embodied systems across five core thematic areas:
- Skill Transfer & Acquisition: Developing intuitive interfaces and learning frameworks that enable robots to acquire, adapt, and refine complex manipulation skills directly from human demonstration.
- Collaborative (Active) Perception: Leveraging multi-robot communication and goal-driven sensor planning to build shared, real-time spatial awareness for mobile autonomous fleets.
- Reinforcement Learning for Optimal Control: Synthesizing data-driven RL methods with model-based control theory to deliver stable, high-performance execution under strict safety and timing constraints.
- Softwarized Infrastructure for Physical AI: Engineering low-latency, softwarized network technologies (e.g., eBPF, P4, openWRT) to seamlessly execute compute-heavy AI workloads at the physical edge.
- Simultaneous Localization & Mapping (SLAM): Pushing the boundaries of robust multi-modal feature fusion and distributed spatial mapping to enable precise, real-time navigation in dynamic or unmapped environments.