Event-Based Imaging: Advancements in Enhancing Visual Perception under Challenging Conditions
Jinshan PanNanjing University of Science and Technology
Personal website
Workshop on Event-Based Vision - IROS 2025
Location
Hangzhou International Expo Center
Important dates
- Workshop date: October 20, 2025
Context
Event-based cameras are bio-inspired visual sensors that mimic the transient pathway of the human visual system, offering
key advantages (e.g., microsecond temporal resolution and high dynamic range) that hold the potential to revolutionize robot
state estimation and image processing. Since the first commercially available event camera in 2008 and the first Workshop on
Event-based Vision at ICRA 2017, the community has witnessed a surge in event-based/-enhanced solutions for robotics and
computer vision. However, the community is facing a chicken-and-egg dilemma: on one hand, the high price of event cameras
stifles the community growth; on the other hand, the absence of large-scale deployment of event-based solutions discourages
mass production of these cameras. To this end, this workshop is dedicated to event-based vision, with a particular focus on its
development in state estimation and image processing.
This workshop builds on the tradition of inviting pioneering figures in the community as speakers, while also serving as a bridge between international/domestic start-ups and academia. It aims to promote discussions on identifying roadblocks that hinder progress in the field and foster collaborative solutions to overcome these barriers. Besides, the first-ever Event-based SLAM Challenge will be held in this workshop. This challenge seeks to benchmark state-of-the-art algorithms, encourage innovation in event-driven/-enhanced approaches, and push the boundaries of what is achievable in real-time ultra-frame-rate state estimation for high-speed robots. As a whole, this workshop will place a strong emphasis on the reproducibility of research findings in real-world scenarios and their tangible impact on advancing robotics technology
Program
| Time | Speaker | Topic/title |
|---|---|---|
| 8:30am-8:40am | Organizers | Welcome talk - Intro of workshop |
| 8:40am-8:55am | SynSense | Industry Talk |
| 8:55am-9:10am | Dvsense | Industry Talk |
| 9:10am-9:30am | - | Mobile robot state estimation based on event camera |
| 9:30am-9:50am | - | Mobile robot state estimation based on event camera |
| 9:50am-10:10am | Jinshan Pan | Event-Based Imaging: Advancements in Enhancing Visual Perception under Challenging Conditions |
| 10:10am-10:30am | Kuk-Jin Yoon | Multi-Modal Fusion in Computer Vision: Leveraging Event Data for Enhanced Object Detection and Scene Understanding |
| 10:30am-10:50am | Yu Lei | Integrating Asynchronous Event Data with New Deep Learning Models: Challenges, Techniques, and Future Directions |
| 10:50am-11:10am | - | Tea Break |
| 11:10am-11:20am | Organizers | Intro of Event-based SLAM Challenge: background, setup |
| 11:20am-11:30am | Organizers | Awards Ceremony |
| 11:30am-11:50am | Winner | Event SLAM Challenge Winner Presentation |
| 11:50am-12:30pm | Panelists | Panel Discussion - TBD |
| 12:30pm | - | End |
Note: All times are in the local time zone of IROS 2025 (Beijing).
Speakers
Nanjing University of Science and Technology
Personal website
Abstact
TBD
Localizing Faster and Sooner: Adventures in Event Cameras and Spiking Neural Networks
Tobias FischerQueensland University of Technology
Personal website
Abstact
Knowing your location has long been fundamental to robotics and has driven major technological advances from industry to academia. Despite significant research advances, critical challenges to enduring deployment remain, including deploying these advances on resource-constrained robots and providing robust localisation capabilities in GPS-denied challenging environments. This talk explores Visual Place Recognition (VPR), which is the ability to recognise previously visited locations using only visual data. I will demonstrate how energy-efficient neuromorphic approaches using event-based cameras and spiking neural networks can provide low-power edge devices with location information with superior energy efficiency, adaptability, and data efficiency.
Organizers
