Topological Twin for Mobility Support Robots

Abstract

Recently, the concept of Cyber-physical Systems has been extended with the technological development on the Internet of Things and Machine Learning. Furthermore, Cyber-physical Systems have been successfully applied to Mobility as a Service (MaaS) and Robotics as a Service (RaaS). Especially, we have to improve the performance of human behavior prediction to deal with the safety of people and the performance of systems simultaneously in both MaaS and RaaS. However, the computational cost is very expensive to estimate and predict human behaviors. In order to reduce the computational cost, we have proposed various methods based on the concept of Topological Twin. In this paper, we discuss the methodology on topological twin for mobility support robots shared in the research on both MaaS and RaaS. First, we explain the concept of topological twin and its related methods on growing neural gas. Next, we show several preliminary experimental results. Finally, we discuss the applicability of topological twin to mobility support robots from the viewpoints of MaaS and RaaS