Geoff Gordon
Papers
15
Total Citations
2,535
H-Index
12
About
Geoff Gordon is a leading figure in artificial intelligence and robotics, renowned for his foundational work on planning under uncertainty. His primary research areas include partially observable Markov decision processes (POMDPs), anytime algorithms, and multi-agent systems. Gordon's most significant contribution is the introduction of Point-Based Value Iteration (PBVI), a breakthrough algorithm that made POMDP planning tractable for real-world problems by approximating solutions through a small set of representative belief points. This seminal work, with over 934 citations, transformed the field and enabled practical applications in robotics and control. He also developed Anytime Dynamic A* (538 citations), a graph-based replanning algorithm that provides bounded suboptimal solutions while reusing previous search efforts, and advanced anytime approximations for large POMDPs (373 citations). His research extends to partially observable stochastic games with common payoffs (160 citations) and learning predictive state representations (93 citations). Gordon's work has had profound impact on robot control, pursuit-evasion problems, and planetary rover monitoring, establishing him as a pioneer in scalable decision-making algorithms for autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1Point-based value iteration: an anytime algorithm for POMDPs934 citations · 2003
- 2Anytime dynamic A*: an anytime, replanning algorithm538 citations · 2005
- 3Anytime Point-Based Approximations for Large POMDPs373 citations · 2006
- 4Anytime search in dynamic graphs246 citations · 2008
- 5
- 6Learning low dimensional predictive representations93 citations · 2004
- 7Policy-contingent abstraction for robust robot control52 citations · 2012
- 8Parallel Stochastic Hill- Climbing with Small Teams38 citations · 2005
- 9Visibility-based Pursuit-evasion with Limited Field of View25 citations · 2006
- 10Efficient Monitoring for Planetary Rovers24 citations · 2003