Two papers have been accepted at the 22nd International Conference on Principles of Knowledge Representation and Reasoning!
1️. "Generalizing Platform-Aware Mission Planning for Infinite-State Timed Transition Systems" by Stefan Panjkovic, Alessandro Cimatti, Andrea Micheli, Stefano Tonetta:
The Platform-Aware Mission Planning (PAMP) problem (introduced in our previous work [1]) formalizes the relationship between an automated temporal planning problem and an execution platform modeled as a Timed Automaton, and consists in finding a valid plan with robustness guarantees w.r.t. a non-deterministic platform. In this paper, we significantly generalize the PAMP problem by considering platforms represented as infinite-state timed transition systems, introducing a new feature to model relations between the planning and platform variables, and generalizing the semantics to cope with unbounded traces. We define a solution method for the resulting generalized PAMP, combining an automated temporal planner and an infinite-state model-checker. Our method is largely more efficient than existing approaches for bounded PAMP problems, despite being strictly more expressive.
[1] Stefan Panjkovic, Alessandro Cimatti, Andrea Micheli and Stefano Tonetta (2025). Platform-Aware Mission Planning. ICAPS 2025. https://arxiv.org/abs/2501.09632
2️. "Counterfactual Scenarios for Automated Planning" by Nicola Gigante, Francesco Leofante, Andrea Micheli:
In this paper, we propose a comprehensive formal framework for counterfactual explanations of automated planners: given a planning problem and an LTLf formula defining desired properties of a plan, counterfactual scenarios identify minimal modifications to the problem such that it admits plans that comply with the formula. We then characterize the computational complexity of generating counterfactual scenarios when different types of changes are allowed. We show that producing counterfactual scenarios is often only as expensive as planning.