Spatio-Temporal Dynamics - Lune

Research Areas

With respect to its basic Research Topics, the contemporary scientific agenda of STeDy addresses several areas within Knowledge Representation and Reasoning (KR) in particular, and reaches out to all sub-disciplines within Artificial Intelligence, Spatial Cognition, and Cognitive Technologies in general. Some core KR areas being addressed include:

Topics of Interest

Main topics of interest include:

Application Areas

Some applications areas of special interest that have also witnessed recent attention include:

Invited Keynotes

Patrick Doherty

Speaker
Patrick Doherty. Professor of Computer Science, Linköping University, Sweden
Leveraging KR Techniques in Autonomous Unmanned Aircraft Systems

Abstract
In this talk I will describe a number of autonomous unmanned aircraft systems we have developed and used as research platforms during the past decade. A major focus of our research effort has been to push AI technologies into fully deployed unmanned aircraft systems. I will show how we have leveraged different knowledge representation systems and integrated them into our platforms. Particular systems that will be discussed are automated temporal planning systems, execution monitoring systems, stream-based reasoning systems and frameworks for cooperative robotics based on delegation. Use of these techniques will be demonstrated in diverse emergency services applications using both single and multiple platform scenarios. Both simulated and actual missions will be shown.

About the Speaker

Patrick Doherty is a Professor of Computer Science at the Department of Computer and Information Sciences (IDA), Linkoping University, Sweden.

He heads the Artificial Intelligence and Integrated Computer Systems Division at IDA. He also serves as director of LinkLab, a research center for future aviation systems, which is a collaborative endeavor between Linköping University and Saab Aerosystems. He is an ECCAI fellow and currently an ECCAI board member. He is also on the board of KR Inc. He has previously served as president of SAIS, the Swedish Artificial Intelligence Society. In addition, he is also an associate editor for the Artificial Intelligence Journal.

His main areas of interest are knowledge representation, automated planning, intelligent autonomous systems and multi-agent systems. A major area of application is with Unmanned Aircraft Systems. His research group has successfully designed and deployed many autonomous Unmanned Aircraft Systems in the past decade. His research group has also won several international competitions pertaining to micro-aerial vehicles and to automated planning. He has over 150 refereed scientific publications in his areas of expertise and has given many keynote and invited talks at leading international conferences such as IJCAI, KR and ECAI. He is also CEO of a new start-up company, UAS Technologies Sweden AB, which designs and markets Micro-Aerial Vehicles.

Christian Freksa

Speaker
Christian Freksa. Professor of Informatics, University of Bremen, Germany
Spatial Computing for Commonsense Reasoning

Abstract
My contribution deals with the relation between natural and formal descriptions of commonsense knowledge and cognitive processes on one hand and their implementations in natural and artificial cognitive systems, on the other hand. The expressive power of general languages and formalisms by far exceeds the structural power of physical systems they describe. In particular, natural and formal languages can describe hypothetical situations and processes which cannot exist in the physical world.

In AI, we use formal languages to abstractly characterize knowledge and commonsense reasoning on the meta-level. The strengths of the formalisms for characterizing cognitive systems may turn out to be a weakness for modeling and implementing them on the object level, where structural constraints of the medium control decisions and actions. I will present some of the structural features of cognitive systems that support cognition and I will discuss spatial computing as an object-level approach to implementing cognitive principles of spatial and temporal processing.

About the Speaker

Christian Freksa is a professor for informatics at the University of Bremen and head of the Cognitive Systems group (CoSy). His research interests focus on knowledge representation, specifically on the representation of spatial and temporal knowledge and on reasoning with qualitative, approximate, incomplete, imprecise, fuzzy, uncertain, and conflicting knowledge. Christian Freksa’s research group employs formal and computational approaches to knowledge representation, designs computer models of spatial environments for simulation studies, and conducts autonomous robotic experiments in the Spatial Environment Laboratory at CoSy using diverse intelligent technologies. He closely cooperates with researchers from other disciplines interested in the investigation of spatial structures, especially cognitive psychology, linguistics, geography, and design.

Christian Freksa directs the Spatial Cognition Research Center SFB/TR 8 funded by the German Research Foundation (DFG), a joint enterprise of the Universities of Bremen and Freiburg, and coordinates the International Quality Network on Spatial Cognition (IQN) established by the German Academic Exchange Service (DAAD). Previously, he coordinated the DFG Spatial Cognition Priority Program and co-coordinated the EU HCM project Spacenet. Christian Freksa is an ECCAI Fellow.

Gérard Ligozat

Speaker
Gérard Ligozat. Emeritus Professor, Paris-Sud University, France
Three decades of Qualitative Spatial and Temporal Reasoning

Abstract
Allen’s 1983 paper on the Interval Algebra introduced constraint propagation techniques to the domain of temporal reasoning. During the following three decades, a significant body of research has been devoted to studying the properties of Allen’s formalism, as well as to proposing a host of spatial and temporal formalisms modeled on it.

Once established that the problem of determining consistency for the Interval Algebra is a NP-complete problem, the search for tractable sub-classes of relations in it, and the analogous question for other formalisms, have occupied a central position in the literature. Basically, two approaches have been used for characterizing tractable subsets: a syntactical approach, and a geometrical approach.

As a consequence of the emergence of many related formalisms, the search for a general framework has also become a timely topic. The introduction of new frameworks has resulted in clarifying the nature of the formalisms, their properties, their expressiveness and the nature of their models.

The relationship of these formalisms with the domain of (finite) CSPs has also been exploited, both on a theoretical level (relating complexity properties to properties of the associated clones) and on a practical level (devising efficient methods for solving consistency problems).

Our talk will start with a presentation of Allen’s formalism from a contemporary perspective, highlighting the points on which important issues have been considered and solved. It will then present some of the main formalisms which have been developed in recent years, and stress the commonalities and differences they exhibit. A general framework based on the notion of a partition scheme will then be discussed and its use for classifying problems and properties will be illustrated. Various extensions to the basic qualitative calculi, such as hybrid calculi and fuzzy calculi will then be presented. Finally, the talk will conclude on a discussion of currently developed approaches as well as long term perspectives for further research.

About the Speaker

Christian Freksa. STeDy 2012
Gérard Ligozat is Professor Emeritus at the Paris-Sud University, France. He held the position of Researcher in Mathematics and Computer Science at the CNRS, and was a Professor of Compuer Science at the University of Paris-Sud, France. He was offered a Professorship in computer science at Adam Mickiewicz University, Poznan, Poland, where he taught Artificial Intelligence in 2008. He has been working on the subject of qualitative temporal and spatial reasoning for the past 25 years.

Pleine Lune Calendrier

Pleine Lune Calendrier

Workshop on Spatio-Temporal Dynamics

ECAI 2020, Lisbon, Portugal
Monday August 16 2020

STeDy 2010 Proceedings
The workshop proceedings have been published as a part of the ECAI 2010 workshop proceedings collection, and also as a part of the Spatial Cognition Research Center (SFB/TR 8) Technical Report Series.

Download: Proceeding, Programme

Invited Keynote
Prof. Antony Galton
University of Exeter, United Kingdom
It is an obvious truth that the possibilities for action and movement are conditioned by the physical spatial environment. In the terminology of J. J. Gibson, these possibilities are defined by the “affordances” of environmental features, and the key to being a successful agent in the physical world is being able to perceive and exploit these affordances. In this talk I want to explore to what extent it is possible to characterise different types of affordance in terms of familiar spatial and temporal logics such as the RCC systems and the interval calculus, e.g., to characterise formally such notions as “container”, “passageway”, “entrance”, and “barrier”, and the types of action or movement that are afforded (or “disafforded”) by environmental features having these properties.
Presentations
The Formalities of Affordance
by Antony Galton, University of Exeter, United Kingdom

The Use of Change Identifiers to Update Footprints of Dot Patterns in Real Time
by Maximillian Dupenois and Antony Galton

Qualitative Spatio-Temporal Data Integration and Conflict Resolution
by Jan Oliver Wallgrün and Frank Dylla

Agent Control by Adaptive Neighborhoods
by Frank Dylla and Arne Kreutzmann

Zooming in on Collective Motion
by Zena Wood and Antony Galton

Spatio-Temporal Abduction for Scenario and Narrative Completion
by Mehul Bhatt and Gregory Flanagan

Improving Solutions of Problems of Motion on Graphs by Redundancy Elimination
by Pavel Surynek and Petr Koupý

On First-Order Propositional Neighborhood Logic: A First Attempt
by Dario Della Monica and Guido Sciavicco

Encoding Spatial Domains with Relational Bayesian Networks
by Valquiria Fenelon, Paulo Santos, Britta Hummel and Fabio Cozman

Knowledge-based Adaptive Thresholding from Qualitative Robot Localisation using Cast Shadows
by Paulo Santos, Valquiria Fenelon Pereira and Hannah Dee

Modeling Motion Event Using QSR
by Rupam Baruah and Shyamanta Hazarika

Program Committee
Christophe Claramunt (Naval Academy Research Institute, France)
Debasis Mitra (Florida Institute of Technology, USA)
Ernest Davis (New York University, USA)
Hans Guesgen (Massey University, New Zealand)
Jochen Renz (Australian National University, Australia)
John Stell (University of Leeds, UK)
Kathleen-Stewart Hornsby (University of Iowa, USA)
Mehul Bhatt (University of Bremen, Germany)
Nico van de Weghe (Ghent University, Belgium)
Paulo E. Santos (Technical University FEI, Brazil)
Shyamanta Hazarika (Tezpur University, India)
Stefan Wolfl (University of Freiburg, Germany)
Thomas Bittner (SUNY at Buffalo, USA)
Organising co-chairs:

Mehul Bhatt (University of Bremen, Germany)
Hans W. Guesgen (Massey University, New Zealand)
Shyamanta Hazarika (Tezpur University, India)