Dr. Pascal Bercher
In July of 2009, I joined the AI Institute at Ulm University, where I am working in the field of Automated Planning, a subfield of Artificial Intelligence (AI). I received my PhD (Dr. rer. nat.) in December 2017 for my thesis Hybrid Planning - From Theory to Practice.
Large parts of my research are motivated by applying AI planning techniques for advanced assistance systems for human users. I am pursuing this goal within the Transregional Collaborative Research Centre SFB/Transregio 62 (SFB-TRR-62) "A companion-technology for cognitive technical systems". I am working in the part project A1, "adaptive planning and decision making". I am also coordinating a transfer project that originated from the previous one, called "Do it yourself, but not alone: Companion-Technology for Home Improvement".
Apart from applying AI planning in practice for the development of assistance systems, my research is concerned with theoretical planning foundations as well as search and heuristics. Please take a look at "Research Interests" (see below) for a more elaborate list of my specific research directions within the field of AI planning.
My research interests can roughly be categorized into two main streams: Foundations of Hierarchical Planning and their application for User-Centered Planning. For a more detailed overview, I want to refer to my dissertation "Hybrid Planning -- From Theory to Practice".
Foundations of Hierarchical Planning
Hierarchical planning refers to a problem class that is centered around the decomposition of abstract tasks until an executable task network is generated. Such a task network consists of only primitive tasks and posseses a linearization that is executable in the initial state. Our institute developed an extension of standard hierarchical planning, which is often referred to as hierarchical task network (HTN) planning, to hybrid planning. Hybrid planning extends HTN planning via concepts known from partial-order causal-link (POCL) planning in the following ways:
- Abstract tasks show preconditions and effects, just like primitive tasks. That way, we can restrict to legal methods. See my ECAI 2016 paper "More than a Name? On Implications of Preconditions and Effects of Compound HTN Planning Tasks" for details.
- Task networks are now called partial plans. Partial plans extend task networks in that they can contain causal links, a concept known from POCL planning.
My research includes the following topics:
- Theoretical foundations of hybrid, HTN, and POCL planning. More specifically, I am interested in the computational complexity of the plan existence problem (of all of these problem classes).
- Search techniques for hybrid planning.
- Heuristics for hybrid, HTN, and POCL planning.
With user-centered planning, we refer to the application of human/computer interaction technology and planning techniques that are centered around human users as a basis to provide automated assistance in a wide variety of possible applications.
User-centered planning capabilities are capabilities that are essential when planning is applied in the real world, in particular when human users are involved. These include:
- Repairing failed plans ("plan repair")
- Finding user-friendly courses of action ("plan linearization")
- Explaining plans to human users ("plan explanation")
- Generating plans together with a human user ("mixed-initiative planning")
We created a prototype system that illustrates the integration of these techniques by providing automated support in setting up a complex home theater. The respective system was demonstrated at AAAI 2015 (see the paper "A Planning-based Assistance System for Setting Up a Home Theater" at AAAI 2015) and described in detail in the ICAPS 2014 paper "Plan, Repair, Execute, Explain - How Planning Helps to Assemble your Home Theater".
Together with Gregor Behnke, Vikas Shivashankar, and Ron Alford, I am organizing the Second ICAPS Workshop for Hierarchical Planning (at ICAPS 2019).
I was the initiator of the very first Workshop for Hierarchical Planning, which was held at ICAPS 2018. I organized it together with Daniel Höller, Susanne Biundo, and Ron Alford. Further information is available on the workshop's webpage: http://icaps18.icaps-conference.org/hierarchicalplanning/
Together with Daniel Höller, I gave the first Tutorial on HTN Planning at ICAPS 2018. Here, you can download the slides.
I was a guest editor for the journal "KI - Künstliche Intelligenz" for the special issue on companion-technologies, which appeared in February 2016. You find the editorial (and other articles that I co-authored in that special issue) in my publication list.
I was in charge of creating a video that promotes our planning-based assistant for setting up a complex home theater. The video further explains the applied scientific technologies for a not necessarily scientific audience.
Program Committee (PC) member at conferences:
- ICAPS: 2019 (senior PC member)
- IJCAI: 2013, 2019
- AAAI: 2015, 2017, 2018 (honored as outstanding PC member), 2019
- IJCAI-ECAI: 2018
Program Committee (PC) member at workshops:
- ICAPS-Hiearchical Planning: 2018, 2019
Reviewer for journals:
- KI - Künstliche Intelligenz
- Theoretical Computer Science (awarded for outstanding reviewing)
Reviewer for conferences (in addition to the PC memberships):
- ICAPS: 2012, 2015, 2016, 2017, 2018
- ECAI: 2016
- AAAI: 2012, 2014
- KI: 2012, 2013
Reviewer for workshops:
- Computer Games Workshop: at IJCAI 2017
- I gave a keynote speech about Companion Systems at the Digital Companion Workshop at MuC (Mensch und Computer; eng.: Human and Computer) 2018 in Dresden, Germany
- KI 2017, where I presented my 2017 IJCAI paper "An admissible HTN planning heuristic" in the "sister conference track".
- In November 2017, I was invited to discuss the potentials and risks of AI in a so-called Junior Science Working Group of a high school in Fulda, Germany, in a plenum discussion together with other experts on the field (Prof. Dr. Gepperth, Prof. Dr. Winzerling, and Dr. Quarch).
I am in charge of managing the following mailing lists:
- Hierarchial-Planning-Experts, from December 2017 until today.
This is a mailing list that I created on my own initiative in December 2017. It is intended for all researchers being interested in the field of hierarchical planning. Its purpose is mainly to give researchers working in that field a convenient possibility to exchange ideas and to come into contact. Currently, it contains around 50 researchers. Please leave me a mail in case you want to subscribe.
- Companion Technology, from December 2017 until January 2019.
This is a mailing list for about 100 researchers that are or were working in the Collaborative Research Centre SFB/TRR 62 A Companion-Technology for Cognitive Technical Systems.
- PLANET, from December 2017 until September 2018.
This is a mailing list for researchers that were working in the PLANET Network of Excellence funded under the Fifth Framework IST Programme of the European Union.
University of Freiburg
While I was still a student of Computer Science at the University of Freiburg, I was a tutor for several lectures. There, being a tutor implied correcting exercises (and exams, but don't tell! :)) and being a lecturer of the respective practice groups. These are the respective lectures:
- Formal Methods and Programming (WS 08/09, Lecture of Cognitive Science)
- Foundations of Artificial Intelligence (SS 08)
- Foundations of Artificial Intelligence (SS 07)
- Computer Science I (WS 05/06)
In SS 2019 I delivered 5 lecture classes in the course Introduction to Computer Science.
In WS 2018/2019 I delivered a lecture on Hierarchical Planning that was conceptualized and delivered by myself.
I was supporting my colleagues in several AI planning and foundational AI lectures. The lectures listed below are those which I was responsible for (which includes creating the exercises and exams as well as being a lecturer of the respective exercise classes).
- Intelligent Planning (SS 18)
- Intelligent Planning (WS 17/18)
- Intelligent Planning (SS 17)
- Introduction to Artificial Intelligence (WS 17/18)
- Introduction to Artificial Intelligence (WS 13/14)
- Introduction to Artificial Intelligence (WS 12/13)
- Introduction to Artificial Intelligence (SS 10)
Seminars organized and supported:
I was a supervisor for one to three seminar participants in all of the following seminars:
- Artificial Intelligence (SS 19) -- this seminar was organized by me
- Artificial Intelligence (SS 18)
- Artificial Intelligence (WS 17/18)
- Artificial Intelligence (SS 17)
- Artificial Intelligence (SS 16)
- Artificial Intelligence (WS 15/16)
- Artificial Intelligence (SS 15)
- Artificial Intelligence (SS 13)
- Artificial Intelligence (SS 12)
- Advances in Artificial Intelligence (WS 17/18)
- Advances in Artificial Intelligence (SS 17)
- Artificial Companions (SS 16)
- Advances in Artificial Intelligence (WS 13/14)
- Advances in Artificial Intelligence (WS 12/13)
- Advances in Artificial Intelligence (SS 12)
I did also supervise several Bachelor, Master, and Diploma theses as well as practicals and research and develpment projects. For a complete list, see the entry "Supervised Theses".
My dissertation won the ICAPS 2019 Best Dissertation Award.
I was selected as Outstanding Program Committee Member at AAAI 2019.
In received an Outstanding Reviewer Award for the (Elsevier) journal Theoretical Computer Science.
The paper A Generic Method to Guide HTN Progression Search with Classical Heuristics by Daniel Höller, me, Gregor Behnke, and Susanne Biundo won the ICAPS 2018 Best Student Paper Award.
The paper Plan and Goal Recognition as HTN Planning by Daniel Höller, Gregor Behnke, me, and Susanne Biundo won the ICTAI 2018 Best Paper Award.
The paper Towards a Companion System Incorporating Human Planning Behavior -- A Qualitative Analysis of Human Strategies by Benedikt Leichtmann, me, Daniel Höller, Gregor Behnke, Susanne Biundo, Verena Nitsch, and Martin Baumann won the TCST 2018 Best Paper Award at the Transdisciplinary Conference on Support Technologies.
Ulm University nominated my dissertation for the GI Best Dissertation Award 2017, a national award (joint with the GI Germany, Switzerland, and Austria) for the best dissertation in the field of Computer Science. The GI (Gesellschaft für Informatik, eng: Society for Computer Science) encourages nominations that make progress in the field of Computer Science or related practical-oriented areas; they should further have some impact on the today's society. The final winners were announced here. All nominated dissertations are described in 10-page abstracts (in German), which will be published in the proceedings "Ausgezeichnete Informatikdissertationen" (eng: awarded Computer Science Dissertations).
The system demo paper "A Planning-based Assistance System for Setting Up a Home Theater" at AAAI 2015 was selected as one of five to be presented at a press conference that was hold during the AAAI conference 2015. They selected only papers that are of interest to the public due to its relevance for today's society.
Since the end of 2016, I am coordinating Ulm's activities of our transferproject "Do it yourself, but not alone: Companion-Technology for Home Improvement" in which our AI institute and the institute of Communications Engineering is working together with Robert Bosch GmbH.
From 2009 to 2017 I was working in the Transregional Collaborative Research Centre SFB/Transregio 62 (SFB-TRR-62) "A companion-technology for cognitive technical systems".
While I was a student at the University of Freiburg, my student research project (similar to Bachelor thesis) as well as my Diploma thesis (similar to Master thesis) -- which resulted in various publications -- were motivated by AVACS (Automatic Verification And Analysis of Complex Systems).
I supervised various theses, projects, and practicals - see list below. Many of those were supervised by me and another colleage (or the other way round). A list of all theses that were written at our institute are listed here.
Diploma and Master Theses:
- On the Computational Complexity of Post-Optimising Partially
Ordered Plans (2018)
- Exploiting Task Decomposition Graphs for Hierarchical Planning (2014)
- Partial-Order Causal-Link Planning mit Präferenzen (2012)
- Algorithms for Identifying and Eliminating Redundant Actions in Partially Ordered Plans (ongoing)
- Extending Partial-Order Causal-Link Heuristics to Cope with Negative Interactions (ongoing)
- A Generic Approach for the Provision of Advice in Modeling Hierarchical Planning Domains (2018)
- Untersuchung von Konsistenz und Adäquatheit von Planungsmodellen (2017)
- The Utility of SAT-Solving for Heuristics in POCL Planning (2016)
- Implementierung und Evaluierung von Suchstrategien für Hybrides Planen (2013)
- Entwicklung einer Methode zur graphischen Darstellung von Planerklärungen (2013)
- Modellierung von Alltagsunterstützung als hybride Planungsdomäne: eine Fallstudie (2012)
Research and Development Projects:
- A Graphical Editor for Drawing Plans with Export to LaTeX tikz (ongoing)
- Investigating the Legality of Hybrid Planning Models as a Basis for Intelligent Modeling Support (2018)
- PANDA3 Visualizer (2016; this project was done in a team of 3 students)
- Effiziente Behandlung von Dekompositionsaxiomen in PANDA2 (2015)
- Experimentiertool für PANDA2: PANDA2 Laboritory (2012)
- On the Computational Complexity of Post-Optimising Partially
I am one of the main developers of PANDA (Planning and Acting in a Network decomposition Architecture). PANDA is a planning system capable of solving hybrid planning problems. As such, it can also solve HTN problems, POCL problems, and classical planning problems.
PANDA1 was developed by my former colleague Dr. Bernd Schattenberg. PANDA2 was mainly developed by me and my former colleage Bastian Seegebarth. We stopped the development and support of both versions. The current version, PANDA3, is mainly developed by my colleagues Gregor Behnke and Daniel Höller. However, the idea of reimplementing the planner was strongly encouraged by me and I was contributing my experience with PANDA2 towards design choices taken for PANDA3. I was also the driving force in making PANDA3 available as open source project and was organizing the prerequisites like legal matters.