REALM: Resource-Aware Edge/On-Device Intelligence for Long-Term Autonomous Mobile Systems

IEEE MASS 2026

Workshop Overview

Autonomous mobile systems, such as robots, drones, intelligent vehicles, and mobile edge devices, are increasingly expected to operate continuously and independently over long periods of time. Unlike cloud-based intelligence, these systems rely on edge and on-device intelligence, where computation, energy, and memory are physically constrained, dynamically varying, and tightly coupled with sensing and actuation.

Recent progress in edge AI, adaptive inference, and foundation models enables more powerful on-device intelligence, but also raises fundamental challenges: How should intelligent capability be provisioned, adapted, and degraded over time when compute and energy resources are limited? How can edge intelligence be co-designed with mobile systems to ensure long-term safety, efficiency, and robustness?

This workshop aims to bring together researchers and practitioners from mobile systems, edge computing, and autonomous intelligence to explore resource-aware edge/on-device intelligence. The focus is on models, systems, and algorithms that explicitly coordinate computation, energy, and intelligent capability to support long-term autonomous operation in dynamic environments.

Topics of Interest

We invite original research contributions, including early-stage and system-oriented work, on topics including but not limited to:

Edge and On-Device Intelligence
  • Edge/on-device AI for mobile and autonomous systems
  • Model compression, quantization, and adaptation for edge deployment
  • Mixture-of-Experts (MoE) and conditional computation on edge devices
  • Adaptive inference, early-exit, and anytime intelligence
Resource-Aware Inference and Scheduling
  • Resource-aware and energy-aware inference mechanisms
  • Joint scheduling of computation, energy, and intelligent capability
  • Capacity-region or budget-based intelligence control
  • Graceful degradation and proportional intelligence supply
Systems and Model Deployment on Edge/Device
  • Systems and middleware for adaptive edge intelligence
  • AI deployment on heterogeneous edge platforms (CPU/GPU/NPU/SSD)
  • Expert caching, migration, and memory hierarchy management
  • Monitoring, profiling, and control of resource usage on devices
Long-Term Autonomous Operation
  • Long-term performance modeling and sustainability of edge intelligence
  • Closed-loop interaction between perception, decision-making, and actuation
  • Energy-aware planning, control, and risk-aware intelligence
  • Cross-layer co-design for long-term autonomous mobile systems
Evaluation, Benchmarks, and Applications
  • Long-duration evaluation and benchmarking of edge intelligence systems
  • Simulation platforms and real-world deployments
  • Applications in robotics, drones, vehicular systems, and mobile sensing
  • Case studies and lessons learned from deployed systems

Workshop Chairs

  • Ning Li, Professor, Harbin Institute of Technology, China
  • Kang Wei, Associate Professor, Southeast University, China

Program Committee Members

  • Dr. Ming Xiao, Full Professor, KTH Royal Institute of Technology, Sweden
  • Dr. Quan Chen, Full Professor, Guangdong University of Technology, China
  • Dr. Tingting Chai, Associate Professor, Harbin Institute of Technology, China
  • Dr. Xin Yuan, Associate Professor, Harbin Institute of Technology, China
  • Dr. Guoqing Chao, Full Professor, Harbin Institute of Technology, China
  • Dr. Yuwen Chen, Associate Professor, Beijing University of Technology, China
  • Dr. Zhaoyu Zhai, Associate Professor, Nanjing Agricultural University, China
  • Dr. Yue Zeng, Associate Professor, Nanjing University of Science and Technology, China

Workshop Schedule

Workshop Paper Submission: July 31, 2026
Workshop Paper Notification: August 15, 2026
Workshop Paper Camera-ready: August 31, 2026

Paper Submissions

Planned format: All submissions should be written in English with a maximum length of 6 single-spaced, double-column pages using 10pt fonts on 8.5 x 11 inch paper, including all figures, tables, and references, in PDF format.

To submit the paper Click here: Paper Submissions.