Search for: All records

In this work, we investigate the application of a multi-objective genetic algorithm to the problem of task allocation in a self-organizing, decentralized, threshold-based swarm. We use a multi-objective genetic algorithm to evolve response thresholds for a simulated swarm engaged in dynamic task allocation problems: two-dimensional and three-dimensional collective tracking. We show that evolved thresholds not only outperform uniformly distributed thresholds and dynamic thresholds but achieve nearly optimal performance on a variety of tracking problem instances (target paths). More importantly, we demonstrate that thresholds evolved for some problem instances generalize to all other problem instances, eliminating the need to evolve new thresholds for each problem instance to be solved. We analyze the properties that allow these paths to serve as universal training instances and show that they are quite natural. After a priori evolution, the response thresholds in our system are static. The problem instances solved by the swarms are highly dynamic, with schedules of task demands that change over time with significant differences in rate and magnitude of change. That the swarm is able to achieve nearly optimal results refutes the common assumption that a swarm must be dynamic to perform well in a dynamic environment. 

Decentralized computational swarms have been used to simulate the workings of insect colonies or hives, often utilizing a response threshold model which underlies agent interaction with dynamic environmental stimuli. Here, we propose a logistics resupply problem in which agents must select from multiple incoming scheduled tasks that generate competing resource demands for workers. This work diverges from previous attempts toward analyzing swarm behaviors by examining relative amounts of stress placed on a multi-agent system in conjunction with two mechanisms of response: variable threshold distribution, or duration level. Further, we demonstrate changes to the general swarm performance’s dependence on paired desynchronization type and schedule design, as the result of varied swarm conditions. 

In this work, we use a multiobjective genetic algorithm to evolve agent response thresholds for a decentralized swarm and demonstrate that swarms with evolved thresholds outperform swarms with thresholds set using other methods. In addition, we provide evidence that the effectiveness of evolved thresholds is due in part to the evolutionary process being able to find, not just good distributions of thresholds for a given task across all agents, but also good combinations of thresholds over all tasks for individual agents. Finally, we show that thresholds evolved for some problem instances can effectively generalize to other problem instances with very different task demands.