abstract
Debate2Rescue is a multi-agent simulation for the dynamic allocation of scarce resources in disaster-response scenarios. When multiple emergencies require the same resource—such as a medical helicopter or heavy machinery—autonomous rescuer agents negotiate by exchanging structured logical arguments rather than numerical bids. The project investigates how argumentation-based negotiation can support transparent and context-sensitive decisions in situations where fixed priority metrics are insufficient.
The system combines BDI agents implemented in Jason/AgentSpeak, logic-based argument evaluation through tuProlog and Arg2P, and a custom Java-based discrete spatial simulator. Rescuer agents perceive the environment, maintain beliefs about emergencies, formulate rescue intentions, and justify their resource requests. A dispatcher coordinates task assignments, while the argumentation engine evaluates conflicts and determines the prevailing request according to domain-specific rules. A dedicated communication protocol introduces performatives such as ARGUE, ATTACK, CONCEDE, and REJECT.
The resulting prototype will provide a visual or command-line monitor and structured event logs, making both spatial coordination and the reasoning behind every resource-allocation decision observable and explainable.
outcomes
