Dagstuhl Seminar 17031
Planning and Robotics
( Jan 15 – Jan 20, 2017 )
(Click in the middle of the image to enlarge)
Permalink
Please use the following short url to reference this page:
https://www.dagstuhl.de/17031
Organizers
- Malik Ghallab (LAAS - Toulouse, FR)
- Nick Hawes (University of Birmingham, GB)
- Daniele Magazzeni (King's College London, GB)
- Brian C. Williams (MIT - Cambridge, US)
Coordinator
- Andrea Orlandini (CNR - Rome, IT)
Contact
- Annette Beyer (for administrative matters)
Schedule
A Planning and Scheduling (P&S) and Robotics were strongly connected in the early days of AI, but became mostly disconnected later on. Indeed, Robotics is one of the most appealing and natural application area for the P&S research community, however such a natural interest seems to not be reflected by advances beyond the state-of-the-art in P&S research in Robotics applications. In light of the accelerated progress and the growth of economic importance of advanced robotics technology, it is essential for the P&S community to respond to the challenges that these applications pose and contribute to the advance of intelligent robotics.
In this perspective, this Dagstuhl Seminar on Planning and Robotics (PlanRob) is organized as a follow up of a scientific initiative started within the P&S research community with a twofold aim. On the one hand, it would constitute a fresh impulse for the P&S community to develop its interests and efforts towards the Robotics research area. On the other hand, it aims at attracting representatives from the Robotics community to discuss their challenges related to planning for autonomous robots (deliberative, reactive, continuous planning and execution etc.) as well as their expectations from the P&S community. Specifically, a PlanRob workshop series (http://pst.istc.cnr.it/planrob/) has been started at ICAPS 2013 in Rome (Italy), followed by a second edition at ICAPS 2014 in Portsmouth (NH, USA), a third edition at ICAPS 2015 in Jerusalem (Israel) and a fourth edition at ICAPS 2016 in London (U.K). The PlanRob workshop editions gathered very good feedback from both the P&S and Robotics communities. This resulted also in the organization of a specific Robotics Track at ICAPS in 2014, 2015 and 2016.
The Dagstuhl Seminar on Planning and Robotics provides a unique opportunity for re- searchers coming from both communities to share their experiences, perspectives and expectations. And, thus, this would foster exciting discussions as well as allow to put goals and ideas on firmer ground. Thus, the general aim of the seminar is to increase the synergy between the two research communities discussing new frontiers and possibly eliciting common research programs. Furthermore, a concrete goal that embodies the spirit of this event is to discuss and define a strategy to create a new generation of researchers who have a deep interest and understanding of the integration of P&S and Robotics. In order to focus the discussion during the seminar, we will ask participants to share thoughts and contributions on synergies and research questions related to techniques and tools, modeling approaches and benchmarks development. In this regard, relevant topics include (but are not limited to): Integration of tasks and motion planning in robotics; Integration of planning to sensing and acting in robotic architectures; Planning for long-term autonomy in robotics; Multi-robot coordination; Human-robot interaction; Adjustable autonomy; Verification and validation of plan-based robot behavior; Challenges and benchmarking.
During the seminar, researchers and people from industry involved in intelligent robotics projects (e.g., funded by European Union or National Science Foundation) will present challenges and opportunities connected to real world applications addressing most of the seminar topics mentioned above. Thus, the seminar will constitute also an opportunity to discuss PlanRob topics when facing realistic situations.
Finally, the seminar with the associated Dagstuhl Report will provide a valuable reference point and will be the best venue for starting the promotion of new initiatives (such as workshops, summer schools, benchmark repository, etc.) for incrementing the synergy between the two communities.
Creative Commons BY 3.0 DE
Malik Ghallab, Nick Hawes, Daniele Magazzeni, Andrea Orlandini, and Brian C. Williams
Automated Planning and Scheduling (P&S) and Robotics were strongly connected in the early days of A.I., but became mostly disconnected later on. Indeed, Robotics is one of the most appealing and natural application area for the P&S research community, however such a natural interest seems to not be reflected by advances beyond the state-of-the-art in P&S research in Robotics applications. In light of the accelerated progress and the growth of economic importance of advanced robotics technology, it is essential for the P&S community to respond to the challenges that these applications pose and contribute to the advance of intelligent robotics.
In this perspective, a Planning and Robotics (PlanRob) initiative within the P&S research community has been recently started with a twofold aim. On the one hand, this initiative would constitute a fresh impulse for the P&S community to develop its interests and efforts towards the Robotics research area. On the other hand, it aims at attracting representatives from the Robotics community to discuss their challenges related to planning for autonomous robots (deliberative, reactive, continuous planning and execution etc.) as well as their expectations from the P&S community.The PlanRob initiative was initiated as a workshop series (http://pst.istc.cnr.it/planrob/) started at the International Conference on Automated Planning and Scheduling (ICAPS) in 2013. The PlanRob workshop editions gathered very good feedback from both the P&S and Robotics communities. And this resulted also in the organisation of a specific Robotics Track at ICAPS since 2014.
The aim of this Dagstuhl Seminar was to reinforce such initiative and increase the synergy between these two research communities. Then, most of the attendees contributed with position statements (whose abstracts are available in this report) to present their major challenges and approaches for addressing them. In general, this involved sharing views, thoughts and contributions across the following main topics:
- Long-term autonomy / Open world planning, providing an overview on issues related to continuous planning for robots with partial information or even incomplete models;
- Knowledge Representation and Reasoning in Planning, with presentations on cognitive features and robot planning;
- Challenges in Industrial, Logistics & Consumer Robotics, providing relevant insights related to deployment of robots in real world scenarios;
- Human-Robot Planning, with a wide overview on planning solutions for dealing with interactions between humans and robots;
- Planning and Execution, discussing issues and challenges related to robust planning and execution for robot control;
- Motion Planning / Hybrid planners, with presentations on integrated solutions for robot control at different levels;
- Reliable and Safe Planning for Robotics, providing an overview of ISO standards for robots and, more in general, investigating the exploitation of formal methods to guarantee reliability in robotic applications;
- Technological Issues in Robot planning/Multi-robot Planning, with statements on technological issues in (multi-) robot solutions.
Each session was animated by (i) an opponent, whose role was to be critical about the position statements and (ii) a moderator, to organise the discussion. Therefore, opponents and moderators have provided a short summary of the session ideas and discussion in dedicated Synthesis Sessions to further foster the discussion.
In addition, two panel sessions have been organised on (i) Evaluation, Benchmarking and Competitions, discussing the experience in RoboCup@Home and the organisation of the new Planning and Execution competition (that will be held in 2017), and (ii) Outreach & Training, discussing about the possible organisation of summer schools and the opening of new scientific networking initiatives (e.g., a COST action).
During the seminar, discussions focused on different issues, challenges, possible solutions and new promising trends over a very wide variety of relevant topics: knowledge representation, modelling issues, the need of incomplete models; cognitive features such as, for instance, learning and goal reasoning; human-aware solutions for flexible human-robot interaction; adaptive solutions for human-robot collaboration; robust execution capable of effectively dealing with failure; integration issues in robotic architecture that, e.g., exploit different kind of models and then perform hybrid reasoning; application of formal methods to provide verification and validation functionalities to guarantee reliable robotic systems; etc. Indeed, addressing the integration of P&S and Robotics for development of intelligent robots entails covering a heterogeneous spectrum of problems, often requiring complex solutions that require a vast set of knowledge and technologies.
During the seminar, there was a very high level of engagement and interaction between the participants, enabling a lively and productive week. The main outcome of the seminar was to share a common understanding of issues and solutions with thorough discussions. And the workshop ended with an open discussion on possible follow ups and possible actions to create further opportunities for fostering synergies and interactions between the two communities.
Andrea Orlandini and Malik Ghallab, Nick Hawes, Daniele Magazzeni, and Brian C. Williams
- Rachid Alami (LAAS - Toulouse, FR) [dblp]
- Iman Awaad (Hochschule Bonn-Rhein-Sieg - St. Augustin, DE) [dblp]
- Roman Bartak (Charles University - Prague, CZ) [dblp]
- Michael Beetz (Universität Bremen, DE) [dblp]
- Ronen I. Brafman (Ben Gurion University - Beer Sheva, IL) [dblp]
- Michael Cashmore (King's College London, GB) [dblp]
- Martin Davies (Guidance Automation Ltd - Leicester, GB) [dblp]
- Minh Do (NASA - Moffett Field, US) [dblp]
- Susan L. Epstein (City University of New York, US) [dblp]
- Alberto Finzi (University of Naples, IT) [dblp]
- Hector Geffner (UPF - Barcelona, ES) [dblp]
- Malik Ghallab (LAAS - Toulouse, FR) [dblp]
- Nick Hawes (University of Birmingham, GB) [dblp]
- Malte Helmert (Universität Basel, CH) [dblp]
- Andreas Hertle (Universität Freiburg, DE) [dblp]
- Joachim Hertzberg (Universität Osnabrück, DE) [dblp]
- Laura M. Hiatt (Naval Research Lab - Washington D.C., US) [dblp]
- Michael W. Hofbaur (Joanneum Research - Klagenfurt/Wörthersee, AT) [dblp]
- Jörg Hoffmann (Universität des Saarlandes, DE) [dblp]
- Felix Ingrand (LAAS - Toulouse, FR) [dblp]
- Luca Iocchi (Sapienza University of Rome, IT) [dblp]
- Gal A. Kaminka (Bar-Ilan University - Ramat Gan, IL) [dblp]
- Erez Karpas (Technion - Haifa, IL) [dblp]
- Oliver E. Kim (University of Birmingham, GB)
- Sven Koenig (USC - Los Angeles, US) [dblp]
- Lars Kunze (University of Birmingham, GB) [dblp]
- Bruno Lacerda (University of Birmingham, GB) [dblp]
- Gerhard Lakemeyer (RWTH Aachen, DE) [dblp]
- Daniele Magazzeni (King's College London, GB) [dblp]
- Lenka Mudrova (University of Birmingham, GB) [dblp]
- Daniele Nardi (Sapienza University of Rome, IT) [dblp]
- Tim Niemueller (RWTH Aachen, DE) [dblp]
- Andrea Orlandini (CNR - Rome, IT) [dblp]
- Amit Kumar Pandey (Aldebaran Robotics - Paris, FR) [dblp]
- Simon Parsons (King's College London, GB) [dblp]
- Ron Petrick (Heriot-Watt University - Edinburgh, GB) [dblp]
- Mark Roberts (Naval Research - Washington, US) [dblp]
- Enrico Scala (Australian National University - Canberra, AU) [dblp]
- Matthias Scheutz (Tufts University - Medford, US) [dblp]
- Reid Simmons (NSF - Arlington, US) [dblp]
- Elizabeth Sklar (King's College London, GB) [dblp]
- Stephen Smith (Carnegie Mellon University - Pittsburgh, US) [dblp]
- Siddharth Srivastava (UTRC - Berkeley, US) [dblp]
- Manuela Veloso (Carnegie Mellon University - Pittsburgh, US) [dblp]
- Brian C. Williams (MIT - Cambridge, US) [dblp]
Classification
- artificial intelligence / robotics
Keywords
- Automated Planning and Scheduling
- Robotics
- Artificial Intelligence
- Plan Execution
- Robust Autonomy
- Goal Reasoning
- Adjustable Autonomy
- Human-Robot Interaction
