ReFLEX Lab

"Cyber-physical systems, such as autonomous robots and self-driving cars, are becoming integral to modern society, with profound economic and societal implications. Our lab envisions leading the advancement of real-time scheduling and assurance techniques that enable safe, reliable, and adaptive robotics and autonomous systems."
— Dr. Steven Xiaotian Dai, Director of ReFLEX lab

Introduction

At ReFLEX Lab, we focus on advancing real-time and embedded systems across topics like scheduling on many-core architectures, harnessing digital twins for performance gains, ensuring reliable timing in robotics and AI, verifying long-lived cyber-physical systems, and innovating hardware for real-time and IoT solutions. Typical target systems of the lab include autonomous driving, avionic and aerospace control systems, robotic warehouse, virtual production and transportation. By blending theory and practice, we aim to push the boundaries of computing and deliver safer, more efficient technologies that can serve both industry and society.

A digital twin is a highly sophisticated virtual replica of a physical asset, process, or system that enables real-time monitoring, analysis, and simulation. By integrating data from sensors and IoT devices, digital twins provide valuable insights into the current state of the physical counterpart, facilitating predictive maintenance, performance optimization, and enhanced decision-making. This technology is widely used across industries such as manufacturing, healthcare, and urban planning to improve operational efficiency and reduce costs. As digital twin technology continues to evolve with advancements in artificial intelligence and machine learning, its ability to simulate complex scenarios with high accuracy makes it an indispensable tool for proactive management and innovation.

Scheduling plays a critical role in optimizing workflows and resource allocation within various organizational contexts. Effective scheduling involves creating timely plans for tasks, appointments, or processes to ensure that objectives are met efficiently while minimizing delays and conflicts. The concept of scheduling goes beyond computer task scheduling. In manufacturing environments, scheduling algorithms help coordinate machine usage and workforce deployment to maximize productivity. In project management, scheduling ensures that milestones are achieved on time by allocating appropriate resources at each phase of the project. Advanced scheduling techniques often incorporate constraints, priorities, and uncertainties to generate robust plans that adapt to changing conditions while maintaining optimal performance.

Our Research

➤ Mixed-criticality scheduling and timing assurance for avionics

This reseach explores advanced scheduling techniques and timing assurance methods tailored for mixed-criticality avionics systems. Given the increasing complexity and safety requirements of modern aircraft, ensuring reliable operation across diverse criticality levels is paramount. The study investigates novel algorithms that optimize task scheduling, enhance predictability, and guarantee timing constraints in environments where high-criticality functions coexist with lower-criticality tasks. The proposed approaches aim to improve system safety, reliability, and efficiency, contributing to the development of robust avionics architectures capable of meeting stringent certification standards.

➤ Multi-core and network scheduling and allocation for autonomous driving

Autonomous driving systems require real-time processing of vast amounts of sensor data, decision-making algorithms, and communication with other vehicles and infrastructure. Efficient scheduling and resource allocation across multiple cores and network components are critical to ensure safety, reliability, and performance. This work explores advanced multi-core processing strategies and network scheduling techniques tailored for autonomous vehicles, aiming to optimize computational efficiency, reduce latency, and enhance system robustness.

➤ Large scale allocation and scheduling for cyber-physical systems

Cyber-physical systems (CPS) are increasingly prevalent in diverse domains such as manufacturing, transportation, healthcare, and energy management. These systems integrate computational elements with physical processes, necessitating efficient allocation and scheduling strategies to ensure optimal performance, safety, and reliability. This paper addresses the challenges of large-scale resource allocation and task scheduling within CPS environments. We propose a scalable framework that leverages advanced optimization techniques and distributed algorithms to manage complex dependencies and real-time constraints. Our approach aims to enhance system throughput, reduce latency, and improve adaptability in dynamic settings. Extensive simulations demonstrate the effectiveness of our methods compared to existing solutions, paving the way for more resilient and efficient cyber-physical infrastructures.

➤ Real-time digital twin and robotics for virtual production

This project explores the integration of advanced digital twin technology with robotics systems to revolutionize virtual production workflows. By creating highly accurate, real-time virtual replicas of physical environments and assets, combined with robotic automation, we aim to enhance efficiency, flexibility, and creative possibilities in film, television, and live events.

➤ Multi-robot safety-aware scheduling for Industry 5.0

As Industry 5.0 advances towards increasingly collaborative human-robot environments, ensuring safety while maintaining efficiency becomes paramount. This study presents a novel multi-robot scheduling framework that integrates safety-awareness into task allocation and sequencing. By leveraging real-time sensor data and predictive analytics, the proposed approach dynamically adapts to operational uncertainties, prioritizing safety without compromising productivity. Simulation results demonstrate improved safety compliance and enhanced operational throughput in complex manufacturing scenarios, paving the way for more resilient and human-centric industrial automation.

Our Team

Lab Lead

Lab Advisors

Research Fellows

  • Dr. Nan Chen (2023-2028), Adaptive Task and Resource Scheduling in Avionics (SCHEME project)
  • Dr. Felix Ulrich-Oltean (2025-2028), Safety-Critical Digital Twin for Avionics (SCHEME project)

PhD Students

  • Modie Al Shakarchi (2025-2027), Real-time Adaptive Digital Twin Robotics for Virtual Production (RADIANT-VP), co-supervisor: Dr. Pengcheng Liu
  • Dr. Zou Jie (2019-2023), Safety-Driven Timing-Predictable and Resource-Efficient Scheduling for Autonomous Systems, co-supervisor: Prof. FREng. John McDermid → Research Fellow, University of York

Academic Visitors

  • Haochun Liang, Real-time containers for Mixed-Criticality Real-Time Systems

MSc Students

  • Xiyu Fang (2025), Randomised DAG generator for evaluting scheduling performance
  • Zilun Zhang (2025), Randomised DAG generator with AI based on small samples
  • Zirui Yuan (2023), Simulation and Optimization of Routerless Networks-on-Chips
  • Zhijian Wang (2021), Priority Assignment Algorithms in Multiprocessor Real-Time Systems with Shared Resources
  • Zixun Yu (2021), Smart Intersection Control with Back-Pressure Algorithms (co-supervised)

UG Students

  • Chase Mo (2024/25), Traffic detection for deriving KPIs
  • Dom Decicco (2024/25), Routerless Network-on-Chip optimization
  • James Sutton (2024/25), Multi-robot scheduling for warehouses
  • Phoebe Russell (2024/25), Traffic control with back-pressure
  • Aron Hogarth (2024/25), Gamification on High Education studying and learning of programming
  • Riko Puusepp (2024/25), AR for accellerating learning of programming
  • Dean Kenny (2020/21), Simulating and improving the scheduling in time-sensitive networks

Contact

Dr. Steven Xiaotian Dai
(xiaotian.dai@york.ac.uk)
Department of Computer Science
University of York, UK
YO10 5GH