IPS2023 - Program

Program

November 7th, 2023

(15min Talk + 5 min Questions)

[11:00 - 12:00] Session 1

Chair: Riccardo De Benedictis

[11:00-11:20] -- Heuristic Planning for Hybrid Dynamical Systems with Constraint Logic Programming, Mikhail Soutchanski and Shaun Mathew.
[11:20-11:40] -- Planning as Theorem Proving with Heuristics, Mikhail Soutchanski and Ryan Young.
[11:40-12:00] -- Investigating Domain-oriented Approaches to Optimization in Timeline-Based Planning, Riccardo De Benedictis, Gloria Beraldo, Gabriella Cortellessa and Amedeo Cesta.

[12:00 - 13:00] Keynote Speaker

Planning, Acting, and Learning: Challenges and (some) Results -- Luciano Serafini and Leonardo Lamanna, Fondazione Bruno Kessler, Trento, Italy.

In the realm of intelligent agents, the process of effective interaction with an unknown environment necessitates the acquisition of an environment model. This model serves as the basis for planning actions to achieve an agent's goals. It typically involves an abstract representation of environmental states and their dynamics. However, when an agent is thrust into an unfamiliar environment, the assumption that it possesses such a model becomes unrealistic.

This talk will delve into the crucial aspects of learning and updating such a symbolic model, a task that also includes acquiring the ability to map the agent's perception of environmental states to abstract representations. Equally important is the challenge of executing these abstract actions efficiently with the available physical resources on the agent's platform.

While previous approaches have focused on the individual learning of these components, often in offline settings, the primary challenge is to address all three simultaneously and in real-time. This demands a continuous process of acting, perceiving, and reasoning. The presentation will illuminate the complexities of this issue and discuss some partial solutions that tackle the ongoing challenge of harmonizing learning, planning, and action.

[13:00 - 14:00] Lunch

[14:00-16:00] Session 2

Chair: Alessandro Umbrico

[14:00-14:20] -- Planning for Temporally Extended Goals in Pure-Past Linear Temporal Logic, Luigi Bonassi, Giuseppe De Giacomo, Marco Favorito, Francesco Fuggitti, Alfonso Emilio Gerevini and Enrico Scala.
[14:20-14:40] -- Action-Failure Resilient Planning, Diego Aineto, Alessandro Gaudenzi, Alfonso Gerevini, Alberto Rovetta, Enrico Scala and Ivan Serina.
[14:40-15:00] -- Planning Safe Collaborative Behaviors through Risk-Aware Heuristic Search, Alex Bonini, Marta Cialdea Mayer, Amedeo Cesta, Andrea Orlandini and Alessandro Umbrico.
[15:00-15:20] -- Taming Discretised PDDL+ through Multiple Discretisations, Matteo Cardellini, Marco Maratea, Francesco Percassi, Enrico Scala and Mauro Vallati.
[15:20-15:40] -- Falsification of Cyber-Physical Systems Using PDDL+ Planning, Diego Aineto, Enrico Scala, Eva Onaindia and Ivan Serina.
[15:40-16:00] -- Tabular Model Learning in Monte Carlo Tree Search, Alberto Castellini, Davide Bragantini, Davide Rossignolo, Federico Segala and Alessandro Farinelli.

[16:00 - 16:30] Cooffee Break

[16:30 - 17:30] Session 3

Chair: Elisa Tosello

[16:30-16:50] -- Learning to Act for Perceiving in Partially Unknown Environments, Leonardo Lamanna, Mohamadreza Faridghasemnia, Alfonso Emilio Gerevini, Alessandro Saetti, Alessandro Saffiotti, Luciano Serafini and Paolo Traverso.
[16:50-17:10] -- Goal Recognition as a Deep Learning Task: the GRNet Approach, Mattia Chiari, Alfonso Emilio Gerevini, Luca Putelli and Ivan Serina.
[17:10-17:30] -- Goal Recognition with Deep Learning and Embedded Representation of State Traces, Mattia Chiari, Alfonso Emilio Gerevini, Francesco Percassi, Luca Putelli, Matteo Olivato and Ivan Serina.

[17:30 - 18:00] Closing Remarks

Chair: Enrico Scala

Page updated

Report abuse