Contact : Aurélie Beynier (aurelie.beynier@lip6.fr) Frédéric Maris (maris@irit.fr)
| GDR IA |  |
Contact : Aurélie Beynier (aurelie.beynier@lip6.fr) Frédéric Maris (maris@irit.fr)
| GDR IA | |
| Accueil — LIP6 - Sorbonne Université, Campus Pierre et Marie Curie, Paris | |
| Salle 24-25/405. |
| Ouverture des 6èmes journées MAFTEC | |
| Salle 24-25/405 |
Andreas Herzig (IRIT-CNRS, Toulouse)
Présentation du groupe de travail MAFTEC et du rapport d'activité 2016-2019.
| Exposés et discussion |
Arthur Queffelec (INRIA/IRISA, Rennes)
Connected Multi-Agent Path Finding (CMAPF).We study a variant of the multi-agent path finding problem (MAPF) in which agents are required to remain connected to each other and to a designated base. This problem has applications in search and rescue missions where the entire execution must be monitored by a human operator. We re-visit the conflict-based search algorithm known for MAPF, and define a variant where conflicts arise from disconnections rather than collisions. We study optimizations, and give experimental results in which we compare our algorithms to the straightforward application of conflict-based search. Aurélie Beynier (LIP6, Sorbonne Université, Paris)
Planification multi agent avec prise en compte d’adversaires non-stationnaires.Les modèles markoviens multiagents décentralisés (DEC-POMDP) offrent un cadre mathématique approprié pour la planification de décisions distribuées dans des environnements partiellement observables et incertains. Toutefois, la plupart des modèles et algorithmes existants supposent que la dynamique du système est stationnaire. Dans les cas où des agents coopératifs doivent se coordonner face à des « agents adversaires » adaptatifs, l’hypothèse de non-stationnarité n’est plus vérifiée. Nous présenterons une extension des processus décisionnels de Markov (mono et multi agents) au cadre non-stationnaire. Dans ce modèle, la dynamique du système est représentée par un ensemble de modes stationnaires entre lesquelles les agents évoluent. Nous décrirons différentes approches de détection des changements de modes et de calcul des politiques. Nous présenterons des applications en patrouille multi-agent et en argumentation stratégique.
| Pause déjeuner | |
| . |
| Exposés et sessions de travail | |
| Salle 24-25/405. |
Elise Perrotin (IRIT, Université Paul Sabatier, Toulouse)
EL-O: Epistemic Logic of Observation.
Joint work with Martin Cooper, Andreas Herzig, Faustine Maffre, Frédéric Maris, Pierre RégnierJulien Vianey (IRIT, Université Paul Sabatier, Toulouse)
Applications of EL-O to simple epistemic planning.
Joint work with Martin Cooper, Andreas Herzig, Frédéric Maris, Elise PerrotinAtelier — Groupes de discussion
Atelier — Groupes de discussion
| Rendez-vous au restaurant Paradis Thaï (132 rue de Tolbiac, Paris 13) |
| Accueil — LIP6 - Sorbonne Université, Campus Pierre et Marie Curie, Paris | |
| Salle 24-25/405 |
| Exposés et discussion |
Julien Vianey (IRIT, Université Paul Sabatier, Toulouse)
Dynamic logic of parallel propositional assignments and its applications to planning.
Joint work with Andreas Herzig and Frédéric MarisAbstract: We introduce a dynamic logic with parallel composition and two kinds of nondeterministic composition, exclusive and inclusive. We show PSPACE completeness of both the model checking and the satisfiability problem and apply our logic to sequential and parallel classical planning where actions have conditional effects. Andreas Herzig (IRIT-CNRS, Toulouse).
Resource Separation in Dynamic Logic of Propositional Assignments.
Joint work with Joseph Boudou and Nicolas TroquardAbstract: We extend dynamic logic of propositional assignments by adding an operator of parallel composition that is inspired by separation logics. In that language, the existence of parallel plans can be expressed by means of the Kleene star operator. We provide an axiomatisation via reduction axioms, thereby establishing decidability. We also prove that the complexity of both the model checking and the satisfiability problem stay in PSPACE.
| Exposé et sessions de travail |
Bruno Zanuttini (GREYC, Université de Caen-Normandie, Caen)
TBA.Atelier — Groupes de discussion
| Pause déjeuner | |
| . |
| Exposé et discussion | |
| Salle 24-25/405 |
Hans van Ditmarsch (LORIA-CNRS, Vandoeuvre-lès-Nancy)
Gossip and Knowledge.Abstract: A well-studied phenomenon in network theory since the 1970s are optimal schedules to distribute information by one-to-one communication between nodes that are connected in the network. One can take these communicative actions to be telephone calls, and protocols to spread information this way are known as gossip protocols or epidemic protocols. A common abstraction is to call the information of each agent its secret, and that the goal of information dissemination is that all agents know all secrets: that is the termination condition of gossip protocols. Following investigations assuming a global scheduler, it is now typically assumed that gossip protocols are distributed in some way, where the only role of the environment is to ensure randomization. Statistical approaches to gossip have taken a large flight since then, wherein network topology is an important parameter. In epistemic gossip protocols, an agent (node) will call another agent not because it is so instructed by a scheduler, or at random, but based on its knowledge or ignorance of the distribution of secrets over the network and of other agents' knowledge or ignorance of that. One such protocol requires that an agent may only call another agent if it does not know the other agent's secret. Epistemic features of gossip protocols may affect their termination, the (order of complexity) expectation of termination, their reachability (what distributions of secrets may occur before all agents know all secrets), and so on. Variations involve agents exchanging telephone numbers in addition to agents exchanging secrets (which results in network expansion), or agents exchanging knowledge about secrets; we may also assume common knowledge of the protocol; further generalizations would involve multi-casts. We present a survey of distributed epistemic gossip protocols.
Atelier — Groupes de discussion
| Rendez-vous au restaurant (à venir...) |
| Accueil — LIP6 - Sorbonne Université, Campus Pierre et Marie Curie, Paris | |
| Salle 24-25/405 |
| Exposés et discussion |
Hans van Ditmarsch (LORIA-CNRS, Vandoeuvre-lès-Nancy)
Everybody Knows that Everybody Knows.
Joint work with Malvin Gattinger, Rahim Ramezanian and Rasoul RamezanianAbstract: A gossip protocol is a procedure for sharing secrets in a network. The basic action in a gossip protocol is a telephone call wherein the caller and the callee exchange all the secrets they know. An agent who knows all secrets is an expert. The usual termination condition is that all agents are experts. Instead, we explore some protocols wherein the termination condition is that all agents know that all agents are experts. We call such agents super-experts. Additionally, we model that agents who already know that all agents are experts, do not make and do not answer calls. We also model that such protocols are common knowledge among the agents. We investigate conditions under which such gossip protocols terminate, both in the synchronous case, where there is a global clock, and in the asynchronous case, where there is not. We show that a protocol with missed calls can terminate faster than the same protocol without missed calls. Cédric Herpson (LIP6, Sorbonne Université)
A Dedicated Testbed for Multi-Agents Problems.Abstract: We present Dedale, an environment for studying multi-agents coordination, learning and decision-making problems under realistic hypotheses. Dedale avoids the 8 fallacies of MAS in which all previous testbed fall and offers open, dynamic, asynchronous and partially observable environments. Highly parametrizable, Dedale allows to tackle either cooperative or competitive exploration, patrolling, pickup and delivery, treasure(s) or agent(s) hunt problems with teams from one to dozens of heterogeneous agents in discrete or continuous environments. The variety of modelable multi-agents problems associated with the possibility to create a peer-to-peer network of Dedale’s environments makes us believe Dedale beeing able to become a unifying environment for both MAS research and teaching communities in their goal to work and evaluate their proposals under real-life hypotheses. Feedback from more than 150 early-users comfort us in this perspective.
| Exposés et discussion |
Exposés libres (proposés durant les journées).
Atelier — Groupes de discussion
| Pause déjeuner | |
| . |
| Sessions de travail | |
| Salle 24-25/405 |
Atelier — Groupes de discussion