Max Robotics — Global Service Robot Directory + US Certification

About

Philipp Hennig is a leading researcher at the intersection of probabilistic numerics, Bayesian inference, and machine learning for robotics and control. His work fundamentally reimagines how numerical algorithms can be treated as probabilistic models, enabling them to quantify uncertainty and make data-efficient decisions. A key contribution is his pioneering approach to Bayesian optimization for reinforcement learning, as demonstrated in his highly cited 2017 paper (111 citations), which introduced a framework for optimally trading off between costly physical experiments and cheaper simulations when tuning control policies. This work has had significant impact on making reinforcement learning practical for real-world robotic systems. Hennig has also made substantial advances in nonparametric regression for control, developing incremental local Gaussian regression (38 citations) that enables robots to learn continuously from streaming data while adapting to non-stationary environments. His research extends to medical applications, including automated treatment planning for radiation therapy using infinite mixture models (34 citations). Through his leadership of the Probabilistic Numerics research group at the University of Tübingen and the Max Planck Institute for Intelligent Systems, Hennig has established a new paradigm where numerical computation itself is treated as a statistical inference problem, influencing fields from robotics to computational science.