Towards Adaptive Environment Perception and Understanding for Autonomous Mobile Robots

Abstract

Autonomous mobile robots (AMRs) are increasingly used in various applications like intralogistics, hospitality and agriculture. Though, their software and abilities are still mainly designed and implemented for specific environments and tasks, impeding an even more widespread deployment of AMRs. Their environment perception and understanding lacks automated adaption to specific operating environments and conditions.In this paper, we propose a software architecture that allows the automated adaption of AMRs’ environment perception and understanding to different operating environments. Further, we provide a concise review of state-of-the-art approaches and technologies needed. We present an exemplary implementation and evaluate the proposed architecture. On this basis, advantages and open research questions of the architecture are derived.