Decentralized Coalition Structure Formation for Interdependent Tasks Allocation

This paper addresses the problem of task allocation among multiple autonomous agents that must accomplish a complex global task. Solutions to the problem have real-world applications in defense, space, disaster management, etc. We solve this problem via agent coalition formation. Multiple coalition formation mechanisms were introduced in prior art, seldom accounting for interdependent tasks. We address this challenge. We introduce an anytime decentralized coalition formation mechanism that enables agents with complementary capabilities to form, autonomously and dynamically, feasible coalition structures that accomplish a global, composite task. The formed structures are incrementally improved via agent replacements to optimize a global utility. We analyze the complexity and show that, although the general problem is NP-hard, our mechanism provides a solution within acceptable time. We present extensive experimental results that illustrate the added value of our approach.