Learning-enhanced Market-based Task Allocation for Disaster Response

Abstract

In this work we propose a market-based task allocation system for disaster response domains. We model the disaster response domain as a team of robots cooperating to extinguish a series of fires that arise due to a disaster. Each fire is associated with a time-decreasing reward for successful mitigation, with the value of the initial reward corresponding to task importance, and the speed of decay of the reward determining the urgency of the task. Deadlines are also associated with each fire, and penalties are assessed if fires are not extinguished by their deadlines. The team of robots aims to maximize summed reward over all emergency tasks, resulting in the lowest overall damage from the series of fires. We first implement a baseline market-based approach to task allocation for disaster response. In the baseline approach the allocation respects task importance and urgency, but agents do a poor job of anticipating future emergencies and are assessed a high number of penalties. We then propose a learning-enhanced market-based approach. Our regression-based technique modifies agents' bids resulting in an allocation that avoids many of the penalties assessed when using the baseline approach; by avoiding penalties and better respecting task importance and urgency the robot team achieves substantially higher overall reward. We illustrate the effectiveness of our approach in a simulated disaster response scenario.