Research of coalition formation in time-bounded multi-agent systems

Coalition formation is an important solution for task allocation in multi-agent systems. But the time constraint of real-time and dynamic systems makes most related research less useful because they pursue full rational agents and optimal coalition, which are nearly impossible in dynamic systems. We aim to present a coalition formation strategy for real-time multi-agent systems. We do not consider the optimization problem much, but using an interlaced backtracking and progression process, our algorithm can find more and more beneficial coalitions. It is timesaving, efficient and easy to implement.