Towards a Lightweight Platform for Human-Robot Interaction in Federated Edge and IoT Environments

Human-Robot Interaction (HRI) applications are becoming increasingly popular in research, especially as they are being used on remote edge devices and Internet-of-Things (IoT) systems. However, these devices often have very limited hardware capabilities, which makes it challenging to run large-scale model training and testing processes efficiently. Moreover, moving and managing different HRI tasks across varied devices creates significant delays and performance issues, ultimately restricting the ability of these systems to interact effectively. In this paper, we introduce a new lightweight platform named HR-D, which is designed to quickly set up and smoothly transfer HRI applications in distributed, federated environments. HR-D uses a container-based, peer-to-peer (P2P) network structure supported by a Distributed Hash Table (DHT), enabling it to scale efficiently and support anywhere from a few dozen to hundreds of thousands of clients across different applications. By incorporating warm-start containers, our platform ensures that containers or clients can be moved or started quickly to meet the real-time needs of specific HRI tasks. Additionally, HR-D offers flexible group management, which helps maintain high resource availability and adaptability on both edge and IoT devices, making it suitable for HRI applications of any size.