Paul Duckworth
University of Leeds, University of Oxford, Oxford Research Group, Science Oxford
Papers
20
Total Citations
513
H-Index
11
About
Paul Duckworth is a researcher whose work sits at the intersection of autonomous robotics, artificial intelligence, and human activity understanding. He is perhaps best known for his contributions to the STRANDS Project, a landmark initiative exploring long-term robot autonomy in real-world environments that has accumulated nearly 200 citations, establishing him as a prominent voice in service robotics. His research spans qualitative spatial reasoning — evidenced by his development of QSRlib, a widely adopted software library for extracting spatial relations from video — and natural language grounding for robotic systems, reflecting a broad commitment to making robots more perceptive and communicative. Duckworth has also made significant contributions to risk-aware planning, developing frameworks that enable mobile robots to navigate partially known or hazardous environments safely using Gaussian processes and Markov Decision Processes. In a notable interdisciplinary turn, he has applied AI expertise to healthcare policy, co-authoring influential studies assessing the automatability of administrative tasks in NHS primary care, work that has directly informed real-world debates about technology and the future of work. Collectively, his portfolio demonstrates a researcher equally comfortable advancing theoretical robotics and translating AI methods into meaningful societal applications.
Research Focus
Key Achievements
Top Papers
- 1The STRANDS Project: Long-Term Autonomy in Everyday Environments196 citations · 2017
- 2
- 3
- 4Natural Language Acquisition and Grounding for Embodied Robotic Systems39 citations · 2017
- 5Unsupervised human activity analysis for intelligent mobile robots31 citations · 2019
- 6Risk-Aware Motion Planning in Partially Known Environments26 citations · 2021
- 7
- 8Inferring Work Task Automatability from AI Expert Evidence18 citations · 2019
- 9
- 10