Georgios Theocharous
Massachusetts Institute of Technology, Michigan State University
Papers
10
Total Citations
329
H-Index
8
About
Georgios Theocharous is a leading researcher in artificial intelligence and robotics, whose work has fundamentally advanced how autonomous systems make decisions under uncertainty. His primary research areas include hierarchical reinforcement learning, partially observable Markov decision processes (POMDPs), and robot navigation. Theocharous pioneered the integration of temporal and spatial abstractions into POMDPs, demonstrating that effective planning does not require considering the entire belief space. His landmark 2003 paper on approximate planning with macro-actions (68 citations) introduced a novel reinforcement learning algorithm over grid-points in belief space, significantly improving computational efficiency. He further revolutionized the field by representing hierarchical POMDPs as dynamic Bayesian networks for multi-scale robot localization (65 citations), enabling robots to navigate complex indoor environments with unprecedented robustness. His 2001 work on learning hierarchical POMDP models (50 citations) established a general framework for modeling partially observable environments using hierarchical hidden Markov models. Collectively, his highly cited contributions have provided the theoretical and practical foundations for scalable, real-world robot navigation systems, making him a pivotal figure in the development of intelligent autonomous agents.
Research Focus
Key Achievements
Top Papers
- 1Approximate Planning in POMDPs with Macro-Actions68 citations · 2003
- 2Representing hierarchical POMDPs as DBNs for multi-scale robot localization65 citations · 2004
- 3
- 4
- 5Rapid Concept Learning for Mobile Robots33 citations · 1998
- 6
- 7
- 8Rapid Concept Learning for Mobile Robots8 citations · 1998
- 9Spatial and Temporal Abstractions in POMDPs Applied to Robot Navigation7 citations · 2005
- 10