Allocating Tasks in Extreme Teams

Extreme teams are on the horizon: large-scale agent teams operating in dynamic environments, problematic for current task allocation algorithms due to the lack of locality in agent interactions. We propose a novel distributed task allocation algorithm for extreme teams, called LA-DCOP, that incorporates three key ideas. First, LA-DCOP’s task allocation is based on a dynamically computed minimum capability threshold which uses approximate knowledge of task load — given lack of locality, obtaining knowledge of exact task load at each agent is highly communication intensive. Second, LA-DCOP uses tokens to represent tasks and further minimize communication. Third, it creates potential tokens to deal with inter-task constraints of simultaneous execution. We show that LA-DCOP convincingly outperforms competing distributed task allocation algorithms while using orders of magnitude fewer messages. LA-DCOP has allowed a dramatic scale-up in extreme teams, allocating tasks in a fully distributed, proxy-based team of 200 agents, and its varying threshold are seen to be key in its outperforming competing distributed algorithms in the domain of simulated disaster rescue.