Skip to main content
The Adaptive Metrics Lab starts scoring what drivers can't see

The Adaptive Metrics Lab starts scoring what drivers can't see

A Slop University paper scores 214 bus stops, car-park boom gates, and school pickup lines on kerbside sightline readiness, and reports a University fund's decision to use the score alone for retrofit eligibility.

A vision-language reading of existing kerbside camera footage now feeds a five-level readiness score, and a University fund uses it to decide where bollards and mirrors go

A parked delivery van, a hedge, the bus itself: kerbside safety programs have always known a sightline can vanish for reasons a static audit never catches, and never had a way to keep score of it as it happens. The Adaptive Metrics Lab now does, reading the camera footage kerbside sites already produce for the instant a waiting pedestrian disappears from an approaching driver’s view.

A paper from the School of Continuous Improvement puts a name to how it works. The Sightline Readiness Maturity Model runs 214 sites — the bus stop, the car-park boom gate, the school pickup line — through five named levels, from no occlusion record at all through to a fix independently reverified inside the same review cycle. Sten Okwuosa, Sunniva Balogun and Tove Ilesanmi built the pipeline from footage sites already generate, and report the levels holding steadiest where a rostered human presence already doubles as the paper trail a higher level requires.

The Adaptive Metrics Lab has spent years learning to ask a site what it can’t see, not just what it can prove it already fixed.

— Senior Research Fellow Sten Okwuosa, Convenor of the Impact Pathway Atlas

“Car parks are going to take longer to close the gap than the school run does, and we’d rather say that plainly now than have the Fund’s first full cycle say it for us,” says Doctoral Candidate Sunniva Balogun.

The Adaptive Metrics Lab has adopted the resulting score as the sole test for the University’s Bollard-and-Mirror Retrofit Fund, with no case-by-case override for a site that falls short. The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.45gm01.