Carlyn Dougherty

Papers

Total Citations

H-Index

About

Carlyn Dougherty is a leading researcher in robot perception and mapping, specializing in open-set 3D scene understanding and task-driven semantic reasoning. Her most notable contribution is the development of **Clio**, a real-time system that constructs open-set 3D scene graphs by leveraging modern tools like SegmentAnything and CLIP. Unlike traditional closed-set maps limited to predefined classes, Clio enables robots to dynamically identify and segment objects based on task-specific goals, even for previously unseen categories. This breakthrough allows autonomous systems to build adaptive, hierarchical representations of their environments in real time, bridging the gap between raw sensor data and actionable semantic knowledge. Dougherty’s work has garnered significant attention, with her 2024 paper on Clio accumulating 35 citations shortly after publication. By integrating class-agnostic segmentation with open-set language understanding, she has fundamentally advanced how robots perceive, reason about, and interact with complex, unstructured spaces. Her research is poised to impact fields from service robotics to autonomous exploration, offering a scalable framework for machines to understand the world as flexibly as humans do.