Dan Griffith

Papers

Total Citations

H-Index

About

Dan Griffith is a leading researcher at the intersection of robotics, computer vision, and 3D scene understanding. His primary contributions lie in developing real-time, open-set perception systems that allow robots to build and reason about their environments in a task-driven manner. Griffith is best known for his pioneering work on **Clio**, a system that creates real-time, open-set 3D scene graphs. This breakthrough leverages modern foundation models like SegmentAnything and CLIP to move beyond traditional closed-set mapping—which was limited to a fixed set of semantic classes—enabling robots to dynamically identify and segment objects relevant to their specific tasks. With his most-cited paper already garnering **35 citations** within its first year, Griffith’s work is rapidly shaping the future of autonomous navigation and manipulation. By fusing task-driven attention with open-set semantic understanding, he is equipping robots with the ability to perceive the world not as a static list of labels, but as a flexible, goal-oriented hierarchy of objects and spaces—a critical step toward truly intelligent, adaptive robotic systems.