I have studied Computer Science at the University of Rostock between 2008 and 2014 and received my Master's degree there. During my studies I have specialised in "Models and Algorithms".
Since January 2014 I have been a PhD student at the Institute of Artificial Intelligence of Ulm University and a member of the graduate school of the Transregional Collaborative Research Centre SFB/TRR 62 "Companion-Technology for Cognitive Technical Systems".
As part of the centre's project A1 "Adaptive Planning and Decision Making" I'm working on mixed-initiative planning and plan explanations. Additionally I'm also concerned with developing new heuristics for hierarchical planning and creating new planning techniques in general.
Since January 2015 I'm also training the University's students participating in the ACM's International Collegiate Programming Contest.
Planners offer a flexible way to assist users in complex tasks (e.g. when having to operate complicated machines or tools) in a competent way. However, most current planning systems are so-called black-box systems. They don't provide the user with any opportunity to influence the plan generation process, nor given any (or not adequate) information and explanation about the process which generated the plan presented to the user.
In my PhD research, I'm mostly concerned with mixed-initiative planning. It aims to overcome the above mentioned problem by directly integrating the user into the planning process, which enables a high degree of adaptability and individuality in the planning process.
To enable this integration, I'm looking at several questions:
- How can the planner interact with the human user in a natural way (i.e. in way humans plan), while simultaneously use efficient planning strategies of modern planners?
- Which decision must be made by the planner (e.g. to avoid a mental overload on the side of the user) and which decisions must be made by the user?
- How can interjections and requests from the user (which were not elicited by questions from the planner) be incorporated into the planning process?
Especially the last question leads to interesting research problems, both of a theoretical and of a practical nature.
In addition to my work on mixed-initiative planning, I'm also concerned with developing new heuristics for hierarchical planning, creating new hierarchical planning techniques altogether, and to investigate the theoretical foundations of hierarchical planning.
I've been a reviewer for the following conferences:
- International Conference on Automated Planning and Scheduling (ICAPS): 2016, 2017
- AAAI Conference on Artificial Intelligence (AAAI): 2017, 2018
- German Conference on Artificial Intelligence (KI): 2016
I've also been a reviewer for the following journal:
- Künstliche Intelligenz - Special Issue on Companion Technologies
I'm regularly involved in the institute's teaching. So far I've taken part in the following lectures and seminars:
- Lecture "Introduction to Computer Science", Summer term 2017
- Lecture "Introduction to Computer Science", Winter term 2016/2017
- Seminar "Advances in Artificial Intelligence", Summer term 2016
- Seminar "Advances in Artificial Intelligence", Winter term 2015/2016
- Lecture "Intelligente Handlungsplanung", Summer term 2015
- Lecture "Introduction to Computer Science", Winter term 2014/2015
I'm also responsible for the programming exercises of the lecture "Introduction to Artificial Intelligence".
Since January 2015 I've been the trainer of the Universities Universities ACM ICPC group. Here I'm training students who will compete in programming contests like the ACM's International Collegiate Programming Contest (ICPC). Since that year we are also participating regularly in both national and international programming contests.
- Gips-Schüle-Rektor-Stipendium 2017
- Scholarships from the German Academic Scholarship Foundation (2009-2014)
- Gold medal at the ACM's ICPC NWERC 2012
- Silver medal at the ACM's ICPC NWERC 2010
I'm currently working in the project A1 "Adaptive Planning and Decision Making" of the Transregional Collaborative Research Centre SFB/TRR 62 "Companion-Technology for Cognitive Technical Systems".
I'm also part of the transfer project "Do it yourself, but not alone" of the SFB/TRR62, being joint project of the Institute of Artificial Intelligence, the Institute of Communications Engineering and the Rober Bosch GmbH.
I've been the supervisor of the following student's theses
- "The Utility of SAT-Solving for Heuristics in POCL Planning", Bachelor, Matthias Englert, 2016
- "Interaktive Handlungsplanung" (Interactive Action planning ), Master, Tobias Schmidke, 2014
- "Entwicklung nutzerunabhängiger Strategien zur Auswahl, Vereinfachung und Zusammenfassung von formalen Planerklärungen" (Developing user-independent strategies for the selection, simplification and consolidation of formal plan explanations), Bachelor, Tim Schneider, 2014
I'm currently the main developer of the hybrid planning system PANDA3.
- email: gregor.behnke(at)uni-ulm.de
- phone: +49 (0)731/50-24113
- fax: +49 (0)731/50-24119
- Gregor Behnke
- Ulm University
- Institute of Artificial Intelligence
- D-89069 Ulm
- building O27, level 4
- room 4405