Adam W. Harley

Stanford University, Carnegie Mellon University

Papers

Total Citations

H-Index

About

Adam W. Harley is a leading researcher in computer vision and robotics, with a focus on enabling machines to understand and interact with dynamic visual environments. His work bridges the gap between perception and language, advancing how robots learn from and track objects in the world. A key contribution is his pioneering approach to object tracking, as demonstrated in his highly cited 2024 paper on "Zero-Shot Open-Vocabulary Tracking with Large Pre-Trained Models" (9 citations), which leverages large pre-trained models to track arbitrary objects without prior category-specific training—a significant leap beyond traditional tracking-by-detection paradigms. Additionally, Harley has made impactful strides in robot learning from human communication. His 2018 work on "Reward Learning from Narrated Demonstrations" (6 citations) introduces a method for robots to infer goals from natural language descriptions paired with demonstrations, moving past rigid programming of poses or images. This research is foundational for more intuitive human-robot interaction. Through these contributions, Harley is shaping a future where robots can parse video streams with open-vocabulary understanding and learn tasks simply through narrated examples.