Decentralized Planning Using Probabilistic Hyperproperties
🤖 AI Summary
This work addresses collaborative planning for multi-agent systems in stochastic, dynamic, and partially observable environments. Methodologically, it introduces a novel modeling and verification framework based on probabilistic hyperproperties—marking the first application of such properties to decentralized planning—to formally specify temporal dependencies across agent trajectories and joint probabilistic objectives. Theoretically, the framework is shown equivalent to a restricted class of Dec-MDPs, and both are proven undecidable, thereby bridging a long-standing theoretical gap between model checking and multi-agent planning. Technically, the approach integrates compositional construction of multi-MDPs, inter-path temporal logic verification, and extended model-checking techniques. Experiments demonstrate substantial improvements in expressive power for collaborative goals and specification flexibility, while enabling reuse of existing distributed planning tools. Overall, this work establishes a new paradigm for hyperproperty-driven multi-agent decision-making.
Application Category
📝 Abstract
Multi-agent planning under stochastic dynamics is usually formalised using decentralized (partially observable) Markov decision processes ( MDPs) and reachability or expected reward specifications. In this paper, we propose a different approach: we use an MDP describing how a single agent operates in an environment and probabilistic hyperproperties to capture desired temporal objectives for a set of decentralized agents operating in the environment. We extend existing approaches for model checking probabilistic hyperproperties to handle temporal formulae relating paths of different agents, thus requiring the self-composition between multiple MDPs. Using several case studies, we demonstrate that our approach provides a flexible and expressive framework to broaden the specification capabilities with respect to existing planning techniques. Additionally, we establish a close connection between a subclass of probabilistic hyperproperties and planning for a particular type of Dec-MDPs, for both of which we show undecidability. This lays the ground for the use of existing decentralized planning tools in the field of probabilistic hyperproperty verification.
Problem
Research questions and friction points this paper is trying to address.
Decentralized multi-agent planning
Probabilistic hyperproperties
Temporal objectives specification
Innovation
Methods, ideas, or system contributions that make the work stand out.
Decentralized probabilistic hyperproperties
Self-composition of multiple MDPs
Temporal formulae for multi-agent paths
🔎 Similar Papers
No similar papers found.
