معرفی
Daniel Höller is a researcher in the Foundations of Artificial Intelligence (FAI) Group at the Department of Computer Science, Saarland University, Germany. He joined the group in January 2020, having previously worked at the Institute of Artificial Intelligence at Ulm University from November 2013 to December 2019. He holds an M.Sc. in Computer Science from Bonn-Rhein-Sieg University, where he studied from 2007 to 2013.
- Ph.D., Computer Science, Ulm University
- M.Sc., Computer Science, Bonn-Rhein-Sieg University (2013)
Daniel Höller's research lies at the intersection of theoretical and practical aspects of AI planning. His primary focus is on Hierarchical Task Network (HTN) planning, where he has made significant contributions to expressivity analysis, solver development, and the use of classical planning heuristics to guide HTN search. He also works on lifted planning, plan repair, plan recognition, and the integration of planning with deep reinforcement learning. His work often involves formal analysis, heuristic development, and the creation of practical planning systems. He is particularly interested in how planning can be made more efficient, reliable, and applicable to real-world problems, including human-aware applications.
His recent publications demonstrate a consistent trend in advancing HTN planning through novel formalisms (e.g., HDDL), sophisticated solving techniques (e.g., progression search, SAT-based approaches), and the development of robust software frameworks (e.g., PANDA, TOAD, LiSAT). His work increasingly bridges planning with learning, exploring how learned models can inform planning and how planning can provide structure for learning. The subfields span formal methods, search algorithms, knowledge representation, and system building.
- ICAPS 2024 Best Dissertation Award for his thesis on hierarchical planning
- SoCS 2024 Best Student Paper Award (co-authored)
- Winner in 4 out of 6 tracks in the 2023 IPC HTN competition
- ICAPS 2018 Best Student Paper Award
- ICTAI 2018 Best Paper Award
- TCTS 2018 Best Paper Award
- Shortlisted for Best Paper at KI 2020
Daniel Höller has been actively involved in teaching and mentoring, having taught courses on Artificial Intelligence and AI Planning at Saarland University, and previously served as a teaching assistant for a wide range of AI and computer science courses at Ulm and Bonn-Rhein-Sieg Universities. He has received funding through his involvement in the Transregional Collaborative Research Center SFB/Transregio 62 at Ulm University. He has organized and contributed to numerous workshops and conferences, demonstrating strong service to the academic community.
Daniel Höller is a core developer of the PANDA planning framework, the TOAD HTN solver, and the LiSAT system for lifted planning. These systems are state-of-the-art tools that implement his research on heuristic search, model transformation, and SAT-based compilation. His work is conducted within the FAI group at Saarland University, a leading research group in automated planning.
حوزههای پژوهشی
