Traffic Zoology
July 7th, 2005Have you ever driven along the motorway and watched the behaviour of other vehicles, seeing them bunch up and stretch out, reacting as one to changes in road conditions? Did you ever wonder if there was some pattern there, some kind of emergent intelligence going on?
I've often wished I could watch a birds-eye view of the M1, following groups of cars as they progress together, seeing how they interact – using each other as shields, leaders or sacrificial lambs. Matthew Frederick Davis Hemming (do his friends call him that, or Matt?) appears to have an excellent vantage point, because his observations have formed a superb article describing traffic zoology. The descriptions are beautiful, if sometimes hard to visualise. I'd be very interested to see someone try and create a visual model of some of these behaviours – something like the bird flocking and herd simulations that were so mesmerising a few years ago.
It would be really cool to take a model motorway and introduce new cars with particular behaviours to see what happens. Perhaps if this research were then formalised it could be used to better plan our road networks? Companies like Cadillac are already experimenting with road trains and autonomous vehicles which automatically follow each other using proximity sensors, peeling down an off ramp when the driver decides to leave the route.
If we had a more detailed understanding of road Zoology and its role in traffic flow, perhaps 'troll' cars could be introduced to busy stretches of road as governors, acting in a particular way to tempt the traffic to order into the informal structures described in the article? It might turn out a bit like when a tour guide takes a group around a museum and people outside the group just drift in and out of earshot, pretending they're not following because they didn't pay for the tour, but nevertheless tracing the same route through the museum as the guide. Drivers wouldn't really acknowledge or even notice they were being guided, but the effect would be the same. So, who wants to write the algorithms? …