Breakthrough in Neuromorphic Computing: ASNA-Flow Accelerator for Real-Time Optical Flow Estimation

Strategic Insights -

Neuromorphic vision systems have gained attention for their potential in edge deployment of optical flow estimation due to their ultralow power and resource efficiency. However, current neuromorphic computing platforms lack specialized architectures optimized for this problem domain. To address this limitation, researchers from Sun Yat-sen University have developed ASNA-Flow, an event-driven asynchronous neuromorphic accelerator specifically tailored for event-based optical flow estimation. ASNA-Flow achieves real-time performance of 104 frames per second (FPS) with ultralow power consumption of 7.9 mW, demonstrating superior energy efficiency of 0.3 pJ per synaptic operation (SOP).

Key Takeaways:

  • ASNA-Flow is a pioneering algorithm-hardware co-design framework for event-based optical flow estimation, featuring a hardware-aware algorithm optimization and systematic data pattern analysis.
  • The accelerator is implemented in TSMC 28-nm CMOS technology and achieves real-time performance of 104 FPS with ultralow power consumption of 7.9 mW.
  • ASNA-Flow demonstrates superior energy efficiency of 0.3 pJ per synaptic operation (SOP), addressing the limitations of current neuromorphic computing platforms.
  • The research establishes the first dedicated neuromorphic computing solution that simultaneously addresses the temporal sparsity, event-driven processing, and energy constraints inherent in optical flow estimation tasks.
  • Financial supporters for this research include the National Natural Science Foundation of China (NSFC) and the Key-Area Research and Development Program of Guangdong Province.
  • The research team includes Shanlin Xiao, Jinghai Wang, Jilong Luo, Bo Li, Lingfeng Zhou, and Zhiyi Yu from Sun Yat-sen University.

Statistics:

  • ASNA-Flow achieves real-time performance of 104 frames per second (FPS).
  • The accelerator consumes ultralow power of 7.9 mW.
  • ASNA-Flow demonstrates superior energy efficiency of 0.3 pJ per synaptic operation (SOP).
  • The research has been peer-reviewed.
  • ASNA-Flow is implemented in TSMC 28-nm CMOS technology.

Sources:

  • "Asna-flow: an Efficient Asynchronous Neuromorphic Accelerator for Real-time Event-based Optical Flow" published in IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2025.
  • NewsRx LLC, "New Findings from Sun Yat-sen University Describe Advances in Technology (Asna-flow: an Efficient Asynchronous Neuromorphic Accelerator for Real-time Event-based Optical Flow)", October 20, 2025.