Human-Centric Goal Reasoning with Ripple-Down Rules
This work addresses scaling issues in goal reasoning for emergency response simulations, but it is incremental as it extends an existing framework with a known learning method.
The paper tackles the problem of scaling goal reasoning in ActorSim by enabling learning from human demonstrations, using Ripple-Down Rules to build decision rules when trainers correct system decisions. The result is that ActorSim can handle an order of magnitude more goals than before.
ActorSim is a goal reasoning framework developed at the Naval Research Laboratory. Originally, all goal reasoning rules were hand-crafted. This work extends ActorSim with the capability of learning by demonstration, that is, when a human trainer disagrees with a decision made by the system, the trainer can take over and show the system the correct decision. The learning component uses Ripple-Down Rules (RDR) to build new decision rules to correctly handle similar cases in the future. The system is demonstrated using the RoboCup Rescue Agent Simulation, which simulates a city-wide disaster, requiring emergency services, including fire, ambulance and police, to be dispatched to different sites to evacuate civilians from dangerous situations. The RDRs are implemented in a scripting language, FrameScript, which is used to mediate between ActorSim and the agent simulator. Using Ripple-Down Rules, ActorSim can scale to an order of magnitude more goals than the previous version.