Human-Centric Goal Reasoning with Ripple-Down Rules

TL;DR

This paper introduces an extension to the ActorSim goal reasoning framework that incorporates learning from human demonstrations using Ripple-Down Rules, enabling scalable and adaptable decision-making in complex disaster simulation scenarios.

Contribution

It presents a novel integration of Ripple-Down Rules for learning in goal reasoning systems, significantly enhancing scalability and adaptability over previous hand-crafted approaches.

Findings

ActorSim with RDR scales to more goals.

System successfully learns from human corrections.

Demonstrated effectiveness in RoboCup Rescue simulation.

Abstract

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.