Mark Donahue

Papers

Total Citations

H-Index

About

Mark Donahue is a researcher working at the intersection of robotics, artificial intelligence, and formal methods, with a primary focus on making learned policies both interpretable and manipulable. His key research areas include imitation learning, Bayesian nonparametrics, and logical automata for planning and control. Donahue’s most significant contribution is his pioneering work on integrating high-level symbolic reasoning with deep learning, enabling autonomous systems to not only learn from expert demonstrations but also to represent their decision-making processes in a human-understandable, logical format. His 2020 paper, "Deep Bayesian Nonparametric Learning of Rules and Plans from Demonstrations with a Learned Automaton Prior," which has garnered 6 citations, introduces a method that models interactions between high-level actions as an automaton, bridging the gap between data-driven learning and formal logic. This work, along with his subsequent research on "Learning and planning with logical automata" (2021), has laid the groundwork for safer and more transparent AI systems. Donahue’s contributions are particularly notable for their potential to enhance human-robot collaboration, where understanding and modifying a robot's learned behavior is critical.