Guest Editors
Dr. Shan Jiang
Email: jiangsh73@mail.sysu.edu.cn
Affiliation: School of Software Engineering, Sun Yat-Sen University, Zhuhai, China
Homepage:
Research Interests: distributed systems, edge large language models
Summary
The field of Artificial Intelligence (AI) is currently witnessing two paradigm-shifting trends. First is the rise of Embodied AI, where intelligent agents (e.g., robots, autonomous vehicles, drones) learn to perceive, reason, and execute tasks through active interaction with the physical environment. Second is the proliferation of Edge Computing, which pushes computation and data storage from centralized clouds to the network edge, closer to where data is generated.
Embodied Edge Intelligence (EEI) represents the critical intersection of these two frontiers. Traditional Embodied AI often relies on heavy cloud-based computing for training and inference. However, this cloud-centric approach struggles to meet the stringent requirements of real-world physical interaction, such as ultra-low latency, high reliability, bandwidth constraints, and data privacy. Conversely, Edge Computing offers the local processing power necessary to make embodied agents autonomous, responsive, and secure. Despite the promise, realizing true EEI presents significant challenges. Specifically, edge devices (e.g., onboard robot chips) are severely limited by power, thermal, and memory constraints, making it difficult to run large-scale Foundation Models. Mechanisms for orchestrating collaboration between heterogeneous embodied agents, edge servers, and the cloud are still immature. Embodied agents must adapt to dynamic physical environments through continuous learning, yet existing edge training algorithms are often insufficient for this task. Bridging the gap between simulation and the real world remains a core hurdle when deploying models on edge hardware.
This Special Issue aims to bring together cutting-edge research from academia and industry to explore the theories, architectures, algorithms, and applications of integrating Embodied AI with Edge Computing, driving the evolution of intelligence from a cloud brain to an embodied edge. We invite high-quality original research papers, comprehensive reviews, and vision papers. Topics of interest include, but are not limited to:
- Cloud-Edge-End collaborative control frameworks for robotics
- Distributed perception and decision-making mechanisms for embodied agents
- Compression and acceleration of Large Models (LLMs/VLMs) for edge devices (Quantization, Pruning, Distillation)
- Resource-constrained multi-modal fusion (Vision, Tactile, Audio, etc.)
- Edge-based Reinforcement Learning (RL) and Imitation Learning
- Multi-robot collaboration and Edge Swarm Intelligence
- Knowledge sharing and semantic communication between embodied agents
- Hardware-aware Sim-to-Real transfer techniques
- Applications and case studies of embodied edge robots
Keywords
embodied AI, edge computing, edge intelligence, robot learning, lightweight foundation models
Login
Register
Submit a Paper
Propose a Special lssue