Real-Time AI Monitoring for Street Safety and Public Trust

Project summary
This project develops and studies AI tools that flag unusual human movement for review. On streets and footpaths, alerts could draw attention to someone who may need assistance when staff cannot watch video continuously.

The working prototype uses human pose estimation to represent movement as a skeleton. It combines movement analysis, risk scoring and checks over time before displaying an alert for human review. Yang Chen demonstrated the system at the CADSI symposium in October 2026 using prerecorded synthetic video. Its performance in live street use has not yet been tested.
The next stage will examine missed events, false alerts and alert delays. Planned participant feedback will explore usefulness, privacy and trust, subject to ethics and site approvals. The system still processes video, so privacy remains an important part of the research.

A future goal is a small device that processes data locally and sends only essential alert information. This would need further development and testing. Radar and Wi-Fi sensing are possible later research directions. The aim is to develop useful monitoring while understanding the conditions under which people would accept it.

Project Team and Collaborators:
Yang Chen, PhD researcher in Information Technology, James Cook University, Cairns. Research and prototype development.
Professor Ickjai Lee, James Cook University. PhD supervision.

Funding Sources to be Acknowledged:
Research support awarded through the James Cook University College of Science and Engineering HDR Competitive Grants scheme, 2026.