Research

I am interested in developing theoretical proofs and computational methods to enable robots and AI to operate safely around people in a way that we can all understand and trust.

Robotic systems promise to revolutionize our homes, cities and roads, but they still struggle with complex physics, changing conditions, and extreme events. At the same time, the excitement around the generative AI boom is tempered by new concerns about potential harms to people from poorly understood interactions for which we don’t yet have robust guardrails. So far, strong safety assurances have eluded both technologies, but I believe each may contain the key to the other.

My work combines control systems, artificial intelligence, and game theory to help robotic and AI systems reason robustly about their own safety despite using inevitably flawed models of the world and other agents.

Specifically, my research focuses on the following areas: 

• Safe robot learning: how can robots acquire new skills, explore unknown environments, and adapt to unexpected situations without risking accidents?

• Safe interaction and corner-case handling: how can robots ensure safety while sharing the space (or the road) with scores of nearby agents, including people, even in the rare but crucial corner cases?

• Human–AI safety: how can increasingly advanced AI systems anticipate and prevent harm to people in open-ended interaction settings like dialogue or content generation?