Edge Multimodal Intelligence for Networked Embodied Robots

Workshop Summary

This workshop focuses on edge multimodal intelligence for networked embodied robots, a compact intersection of pattern recognition, machine learning, IoT, and robotics. It targets robots that must perceive, reason, and interact in real time under limited computing, energy, privacy, and connectivity conditions. The workshop will discuss lightweight multimodal foundation models, on-device and edge inference, federated/continual learning for robot agents, real-time perception-action loops, and trustworthy embodied AI. The goal is to connect PRAI communities in robot perception, IoT applications, and privacy-aware machine learning, and to promote deployable robotic AI systems with measurable perception accuracy, interaction quality, robustness, and resource efficiency.

中文摘要：
本研讨会聚焦“面向网络化具身机器人的边缘多模态智能”，定位于模式识别、机器学习、物联网与机器人应用的交叉方向。研讨会重点讨论具身机器人在受限算力、能耗、隐私和网络条件下，如何实现实时感知、推理与交互。核心内容包括轻量化多模态基础模型、端侧/边缘推理、机器人智能体的联邦学习与持续学习、实时感知-决策闭环，以及可信具身智能。本研讨会旨在对接 PRAI 的机器人感知、物联网应用和隐私计算/联邦学习等主题，推动可部署、可评测、可复现的具身智能系统研究。

Topics of Interest

• 1. Multimodal Perception and Scene Understanding for Embodied Robots (具身机器人的多模态感知与场景理解)
• 2. Lightweight Multimodal Foundation Models for On-device and Edge Robotic Agents (端侧/边缘机器人的轻量化多模态基础模型)
• 3. Federated, Continual, and Collaborative Learning for Networked Robots (网络化机器人的联邦学习、持续学习与协同学习)
• 4. Real-time Perception-Action Loops and Human-Robot Interaction (实时感知-决策闭环与人机交互)
• 5. Edge-Cloud Co-inference and Resource-aware Robot Intelligence (边云协同推理与资源感知机器人智能)
• 6. Trustworthy and Robust Embodied AI Systems (可信与鲁棒具身智能系统)

Estimated Number of Papers: 6-8 papers

Workshop Chairs