Hugo Buurmeijer

Papers

Total Citations

H-Index

About

Hugo Buurmeijer is a researcher advancing the frontier of safe decision-making under uncertainty, with a primary focus on constrained planning in partially observable environments. His key contributions lie in the development of scalable algorithms for Constrained Partially Observable Markov Decision Processes (CPOMDPs), a framework that generalizes traditional planning by requiring agents to maximize reward while satisfying hard cost constraints—critical for safety-critical applications like autonomous systems and robotics. His most-cited work, "Constrained Hierarchical Monte Carlo Belief-State Planning" (2024), tackles the formidable challenge of online CPOMDP planning in large or continuous state spaces, introducing a hierarchical Monte Carlo approach that efficiently navigates belief states to produce optimal, constraint-satisfying plans. Though early in its impact, this work has already garnered attention for addressing a notoriously difficult problem in AI safety. Buurmeijer’s research bridges theoretical rigor and practical scalability, offering a pathway to deploy reliable autonomous agents in complex, uncertain environments. His achievements underscore a commitment to making principled decision-making viable for real-world constraints, positioning him as an emerging voice in the intersection of planning, reinforcement learning, and safe AI.