Punishment Policy Adaptation in a Road Junction Regulation System

摘要

This paper studies the problem of adapting punishment policies in traffic scenarios. It focuses on a two-road junction scenario simulated by means of Simma, a Multi-Agent Based Simulation Tool. Adaptation is based on an adaptive neuro-fuzzy inference system (ANFIS) together with a hybrid learning algorithm (HLA). Basic punishment policy is provided through a knowledge base that specifies the conditions that must hold in order to assign different punishments. The aim of this paper is to show how the ANFIS system can improve this policy unsupervisedly.