PS-17Safe & Smart Communities
Vision Zero
AI Road-Risk and Near-Miss Intelligence for Safer Streets
SDG 3SDG 11
The challenge
Road-safety interventions often rely on reported crashes, which reveal dangerous locations only after harm has occurred. Near misses, speeding patterns, unsafe crossings and recurring conflict points can provide earlier evidence. Create an AI-based road-risk analytics system that uses sample traffic video, road geometry, speed or incident data to identify conflict patterns and rank locations for engineering or enforcement review. The system should protect privacy, avoid face identification and present risk as decision support rather than automatic fault assignment.
What a working submission needs
Build a prototype that processes recorded/simulated traffic data, detects vehicles/pedestrians, derives at least one near-miss or conflict indicator, maps risk hotspots and generates a transparent recommendation report.
Stretch goals
short-term risk prediction, safer route suggestions, signal-timing analysis or before/after intervention comparison.
Optional. A working core counts for more than feature count.Suggested stack
Web/mobile application, AI/ML models, a mapping or dashboard interface, and open APIs as appropriate to the problem.
Scope and responsibility
Healthcare, safety and public-sector challenges are decision-support tools, not autonomous clinical, legal, emergency-command or enforcement systems. Name your prototype’s limitations in the demo. Judging weighs a working core, justified use of AI, technical quality, real-world impact and responsible design above feature count.
Also in Safe & Smart Communities