Tobias Johannink

Siemens (Germany), Hamburg University of Technology

Papers

Total Citations

H-Index

About

Tobias Johannink is a robotics researcher whose work bridges the gap between classical control theory and modern machine learning. His primary research areas include robot control, reinforcement learning, and autonomous navigation, with a particular focus on solving real-world manipulation and locomotion challenges. Johannink’s most impactful contribution is his work on **residual reinforcement learning**, introduced in his highly cited 2019 paper (45 citations). This approach elegantly combines conventional feedback controllers with reinforcement learning, allowing robots to handle difficult contact-rich tasks—such as those involving friction and deformable objects—that pure model-based methods struggle to solve. By using a base controller to handle the known dynamics and a learned residual policy to compensate for unmodeled effects, his method significantly improves sample efficiency and robustness. Additionally, Johannink has contributed to **micro underwater robotics**, developing an integrated navigation and control system using radio-frequency localization for autonomous vehicles too small for traditional sensors. His work is notable for its practical, systems-level thinking, demonstrating how to deploy advanced algorithms on resource-constrained hardware. With a growing citation record and a focus on deployable solutions, Johannink is establishing himself as a key figure in making reinforcement learning practical for real-world robotic systems.