Implementação de agentes éticos na resolução de conflitos envolvendo veículos autônomos em um cruzamento de trânsito

Abstract

This work presents the development and validation of a multi-agent system (MAS) capable of performing ethical reasoning in decisions involving autonomous vehicles at urban intersections. The investigated problem arises from the need to incorporate moral principles into autonomous systems that interact with human lives, ensuring compliance with ethical guidelines and social acceptance. The general objective consists of modeling and implementing a MAS following the ethical agent architecture proposed by Alves, extending it through the application of Galtung’s Triangle for conflict formalization and MODD rules for decision-making autonomy delimitation. The system was implemented using three BDI agents in the MASPY framework: the AV Agent responsible for autonomous driving, the Ethical Agent acting as Governor Module for monitoring and ethical utility calculation based on the German Ethics Code, and the HITL Agent representing human supervision. Communication between agents uses an adaptation of the Contract Net protocol. Experiments validated the system through urban intersection scenarios, confirming that the system correctly identifies situations resolvable autonomously and those requiring transfer to human supervision, respecting established ethical guidelines.