Going beyond BDI for agent-based simulation

John Bruntse Larsen

doi:10.1080/24751839.2019.1620024

John Bruntse Larsen

Open Access

https://doi.org/10.1080/24751839.2019.1620024

Copy DOI

Abstract

ABSTRACTResearch in multi-agent systems has resulted in agent programming languages and logics that are used as a foundation for engineering multi-agent systems. Research includes reusable agent programming platforms for engineering agent systems with environments, agent behaviour, communication protocols and social behaviour, and work on verification. Agent-based simulation is an approach for simulation that also uses the notion of agents. Although agent programming languages and logics are much less used in agent-based simulation, there are successful examples with agents designed according to the BDI paradigm, and work that combines agent-based simulation platforms with agent programming platforms. This paper analyses and evaluates benefits of using agent programming languages and logics for agent-based simulation. In particular, the paper considers the use of agent programming languages and logics in a case study of simulating emergency care units.

Highlights

Agent-Oriented Programming (AOP) is a programming paradigm where programs are composed of agents
To illustrate the need for going beyond BDI models in agent-based simulation, we describe, analyse and evaluate a simulation of emergency care units we have implemented by using the BDI framework of the ABS platform GAMA
The simulation only covers a small part of the actual treatment process in an emergency care unit, we can evaluate it in terms of the criteria proposed in the introduction and discuss the general limitations of the BDI-based approach

Summary

Introduction

Agent-Oriented Programming (AOP) is a programming paradigm where programs are composed of agents. Agent programming languages are programming languages that are designed for development of multi-agent systems with AOP. Examples of platforms that use agent programming languages include Agent-0 (Shoham, 1993), 3APL (Hindriks, De Boer, Van Der Hoek, & Meyer, 1999), 2APL (Dastani, 2008), Jason (Bordini, Hübner, & Wooldridge, 2007), JACK (Busetta, Ronnquist, Hodgson, & Lucas, 1999; Winikoff, 2005) and GOAL (Hindriks, 2009). Logic provides a theoretic framework for specification and verification of agent programs. Work in the AOP community has resulted in frameworks and meta-models for Multi-Agent-Oriented Programming (MAOP)

Objectives

Results

Discussion

Conclusion