Allocating tasks in extreme teams

Abstract

Extreme teams, large-scale agent teams operating in dynamic environments, are on the horizon. Such environments are 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 overall task load. 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, allowing a dramatic scale-up in extreme teams, upto a fully distributed, proxybased team of 200 agents. Varying threshold are seen as a key to outperforming competing distributed algorithms in the domain of simulated disaster rescue.

Human–agent collaboration for disaster response

“…Since then, a number of algorithms and simulation platforms have been developed to solve the computational challenges involved. For example, algorithms have been developed to efficiently allocate emergency responders to rescue tasks (e.g., to rescue civilians, extinguish fires, or unblock roads) for (i) decentralised coordination: where emergency responders need to choose their actions based on local knowledge [12,43], (ii) coordination by a central authority: where a command centre is able to choose actions (against potentially adversarial agents) for all the members of the team given complete knowledge of the system [26,30,50,58], and (iii) coalition formation: where sub-teams can perform tasks with different levels of efficiency, as defined by the synergies between their capabilities (e.g., when two policemen help rescue a civilian from rubble they would be less effective than a fire and rescue officer and a medic) [45]. Similar to [26,30,50], in our work, we adopt a centralised approach to the coordination problem to and additionally consider the uncertainty in the environment (see more details in Sect.…”

“…Previous agent-based models for team coordination in disaster response typically assume deterministic task executions and environments [45,50]. However, in order to evaluate agentguided coordination in a real-world environment, it is important to consider uncertainties due to player behaviours and the environment (as discussed in the previous section).…”

Dynamic Assignment of Geospatial-Temporal Macro Tasks to Agents under Human Strategic Decisions for Centralized Scheduling in Multi-Agent Systems

“…They assist agents to achieve their objectives and to maximize the benefits of the system. There are works that address the task-scheduling problem in multi-agent systems [14,15], multi-robot systems [16], disaster-emergency teams [17], robocup rescue simulations [18], and strategic decision making for coordinating actions of a USAR team [19].…”

Capability-based task allocation in emergency-response environments: A coalition formation approach

“…In the recent works performed on task allocation in cooperative multiagent environments, attempts have been made to simulate the problem as a distributed constraint optimization problem (DCOP) or a generalized assignment problem (GAP), and to convert the allocation issue into a known problem in the above fields and solve it by heuristic methods [4,5]. These studies emphasized the GAP and restrictions on the allocation and task implementation for heterogeneous tasks, and synergism was disregarded.…”