My research is in the intersection of perception and decision making for embodied systems. How does natural intelligence use perception to build a representation of the world that lets an agent choose actions to act safely, reliably and efficiently? How should an agent act to gather data to let it learn more effectively? To test our understanding of how to act in the real world, my students and I build robots of all kinds, manipulators, ground rovers and UAVs that are designed to learn about and act in an uncertain and changing world.