Mixed-initiative distributed simultaneous (MDS) planning

Abstract

In many real-world situations, a problem emerges suddenly (e.g., Cyber attacks may take just milliseconds), while the response processes may last much longer (e.g., network system restoration after a Cyber attack may take hours or days). Presently, a fully automated response is impractical, due to complex dynamics and uncertainties, as well as social and political requirements (e.g., for humans to be “in the loop”). Thus, a viable approach is to augment human capabilities to solve the problem, but this results in a subsequent complication: the technical skills required are not all residing in one place and a network of dispersed experts must plan and carry out an effective response together. This paper aims to define and investigate a special challenge, named Mixed-initiative Distributed Simultaneous (MDS) planning, arising from complex real-world problems. A framework named We-Plan, its multi-agent based system architecture and a software prototype have been designed and developed to address MDS planning. We present the appropriate background and a definition of the challenge, a brief literature review, an architecture design, and a software prototype implementation featuring a step-by-step walkthrough demonstrating how collaborative processes and multi-agent technologies can be employed to effectively carry out MDS planning. The paper does not delve into algorithmic details as a partial solution, but aims to present a viable and practical framework for addressing the aforementioned challenges.