Multi-agent Systems Meet Aggregate Programming: Towards a Notion of Aggregate Plan

Recent works foster the idea of engineering distributed situated systems by taking an aggregate stance: design and development are better conducted by abstracting away from individuals' details, directly programming overall system behaviour instead. Concerns like interaction protocols, self-organisation, adaptation, and large-scaleness, are automatically hidden under the hood of the platform supporting aggregate programming. This paper aims at bridging the apparently significant gap between this idea and agent autonomy, paving the way towards an aggregate computing approach for multi-agent systems. Specifically, we introduce and analyse the idea of "aggregate plan": a collective plan to be played by a dynamic team of cooperating agents.