Keiji Kanazawa

Brown University, University of California, Berkeley

Papers

Total Citations

H-Index

About

Keiji Kanazawa’s research lies at the intersection of decision theory, control, and robotics, with a focus on how agents reason about change under uncertainty. His work explores the deep connections between observability and controllability, arguing that insights from control theory should fundamentally shape how we approach planning in artificial intelligence. In his 2002 paper, “Prediction, observation and estimation in planning and control,” Kanazawa examines how reasoning about dynamic environments is central to effective planning and control, bridging gaps between estimation theory and decision-making. His earlier work, “Sensible decisions: toward a theory of decision-theoretic information invariants,” proposes a novel framework for bounded rational decision-making, demonstrating how optimal decisions in sensory robotics can be preserved under transformations of the decision rule—a key insight for designing robust, resource-limited autonomous systems. Though his citation counts are modest, Kanazawa’s contributions are conceptually rich, offering foundational ideas for researchers working on decision-theoretic planning, sensor-based control, and the theoretical underpinnings of intelligent action in uncertain environments.