Logo

Applied Spiking Architectures: From Dynamic Visual Sensing to Transparent Physical Actuation


Developing an end-to-end, biologically inspired computing stack that pairs event-based vision with explainable Spiking Neural Networks (SNNs) to enable robust, power-efficient, and transparent perception for physical autonomous systems operating in dynamic real-world environments.

Date: 2024 - 2027

Persons participating in the project:

  • PIs: Dr. Francisco Cruz, A/Prof. Leo Wu
  • Associates: Oltan Sevinc
  • Corresponding contact: m.sevinc@unsw.edu.au

Research areas:
  • Spiking Neural Networks (SNNs)
  • Event-based Vision
  • Neuromorphic Computing
  • Explainable AI (XAI)
  • Computer Vision
  • Embodied AI and Autonomous Systems

Description:
Traditional computer vision architectures struggle when deployed on physical edge devices operating in unpredictable environments, often suffering from high latency, massive power constraints, and failure under extreme lighting variations. This project aims to close the loop on full-stack neuromorphic engineering by designing a robust and interpretable "sensor-to-action" pipeline.

The research spans three core pillars:
  • Robust Event-Driven Sensing: Leveraging the high temporal resolution and high dynamic range of event cameras to maintain perceptual robustness under volatile, real-world lighting conditions.
  • Transparent Spiking Computation: Utilizing Spiking Neural Networks (SNNs) for energy-efficient, asynchronous processing, while developing novel explainability frameworks (such as Class Activation Maps tailored for SNNs) to untangle the "black box" nature of spiking architectures.
  • Physical Edge Actuation: Deploying these robust, interpretable models onto physical hardware platforms to validate end-to-end event-driven intelligence in real-time robotic or autonomous scenarios.

Media:
Additional images/video









Selected Publications Web
Oltan Sevinc, M., Wu, L., & Cruz, F. (2025). Towards Closing the Domain Gap with Event Cameras. Proceedings of the Australasian Conference on Robotics and Automation (ACRA), 2025.