Penalization Framework For Autonomous Agents Using Answer Set Programming

TL;DR

This paper introduces a framework for enforcing penalties on autonomous agents using answer set programming, enabling agents to select compliant plans while considering penalties, thus ensuring adherence to policies in dynamic environments.

Contribution

It proposes a novel framework and algorithm for representing and reasoning about penalties in agent plans within answer set programming.

Findings

Agents can choose plans with minimal penalties under the framework.

The framework effectively reprimands non-compliant agents.

Agents prioritize emergency goals over penalties when necessary.

Abstract

This paper presents a framework for enforcing penalties on intelligent agents that do not comply with authorization or obligation policies in a changing environment. A framework is proposed to represent and reason about penalties in plans, and an algorithm is proposed to penalize an agent's actions based on their level of compliance with respect to authorization and obligation policies. Being aware of penalties an agent can choose a plan with a minimal total penalty, unless there is an emergency goal like saving a human's life. The paper concludes that this framework can reprimand insubordinate agents.