Roderick Bloem

Institut für Informationsverarbeitung, University of Bremen, Graz University of Technology

Papers

Total Citations

H-Index

About

Roderick Bloem is a prominent computer science researcher whose work bridges formal methods, artificial intelligence, and autonomous systems. His most significant contributions lie at the intersection of reinforcement learning and safety verification, where he has pioneered the development of probabilistic shielding techniques for safe reinforcement learning. His highly cited 2020 paper on probabilistic shields, which has accumulated 52 citations, introduced a rigorous framework for constructing safety shields over Markov decision processes, enabling reinforcement learning agents to operate safely under uncertainty — a critical challenge in deploying AI in real-world environments. Bloem's research extends into robotics and autonomous systems, as demonstrated by his 2015 work on synthesizing cooperative reactive mission plans, which showed how formal specification techniques could guarantee correct robot controller behavior. This synthesis-from-specification approach reflects a consistent theme across his career: using mathematical rigor to provide provable guarantees in complex, uncertain environments. His earlier work in formal verification of control software further establishes his long-standing commitment to safety-critical systems. Collectively, Bloem's research has shaped how the AI and formal methods communities think about integrating safety constraints into learning systems, making his contributions increasingly relevant as autonomous systems become more widely deployed.