Welcome to the UCLA Robot Intelligence Lab (URIL)!

Our goal is to build robots that assist people in everyday environments. Today's robots still struggle on three fronts: learning from demonstrations that are scarce and uneven in quality, acting on foundation models that are fluent but not yet dependable, and working well with the people around them. Our research tackles these challenges through three pillars: quality data for robot learning, reliable robot foundation models, and human-centered autonomy.

Quality Data for Robot Learning Robot learning is bottlenecked by scarce, uneven demonstration data. We build accessible collection interfaces, data-quality metrics, and simulated experience to make every demonstration count. Reliable Robot Foundation Models Foundation models produce fluent plans that robots cannot always trust or execute. We ground language and vision-language models in perception and action, adding structure and formal guarantees. Human-Centered Robot Autonomy Autonomy that ignores its users can frustrate them and erode their skills. We design robots and AI systems that infer intent, share control, and make the people around them more capable.

Updates