A Brief Guide to Multi-Objective Reinforcement Learning and Planning

Conor F. Hayes (University of Galway), Roxana Rădulescu (Vrije Universiteit Brussel), Eugenio Bargiacchi (Vrije Universiteit Brussel), Johan Källström (Linköping University), Matthew Macfarlane (University of Amsterdam), Mathieu Reymond (Vrije Universiteit Brussel), Timothy Verstraeten (Vrije Universiteit Brussel), Luisa M. Zintgraf (University of Oxford), Richard Dazeley (Deakin University), Fredrik Heintz (Linköping University), Enda Howley (University of Galway), Athirai A. Irissappane (Amazon), Patrick Mannion (University of Galway), Ann Nowé (Vrije Universiteit Brussel), Gabriel Ramos (Universidade of Vale do Rio dos Sinos), Marcello Restelli (Politecnico di Milano), Peter Vamplew (Federation University), Diederik M. Roijers (City of Amsterdam)

Abstract

Real-world sequential decision-making tasks are usually complex, and require trade-offs between multiple-often conflicting-objectives. However, the majority of research in reinforcement learning (RL) and decision-theoretic planning assumes a single objective, or that multiple objectives can be handled via a predefined weighted sum over the objectives. Such approaches may oversimplify the underlying problem, and produce suboptimal results. This extended abstract outlines the limitations of using a semi-blind iterative process to solve multi-objective decision making problems. Our extended paper [4], serves as a guide for the application of explicitly multi-objective methods to difficult problems.