Home>Cars>Microsoft’s Plan For Car Crashes

Microsoft’s Plan For Car Crashes

Over in the USA, Microsoft and the city of Bellevue, Washington, have a futuristic plan to prevent cyclists and pedestrians from being hit by cars. The project also analyse traffic patterns to predict when and where accidents are mostly likely to occur, and prevent them before they happen.

“This is like ‘Minority Report,’ ” Bellevue senior transportation planner Franz Loewenherz said, referring to the 2002 sci-fi movie in which an elite police unit is able to peer into the future to see crimes that haven’t been committed yet. (In fairness, the movie takes a pretty dim view of arresting people based on such predictions, but doesn’t necessarily condemn the basic idea of trying to anticipate what events are going to happen.)

Rather than relying on gifted clairvoyants to look into the future, Microsoft plans to use video footage and predictive algorithms.

The technology would take video from Bellevue’s existing traffic cameras and run it through special software to determine what combinations of cars, bikes, and pedestrians are using a street or pavement at any moment in time. The software would automatically determine the direction of travel and rate of speed for each, then look for near misses in which a car almost hit a bicycle or a pedestrian.

By crunching all this data, the software would predict which junctions are most likely to see accidents in the future, and would suggest changes such as adjusting traffic light timings in order to lower the probability of a collision.

“If we nail it, this is huge,” Victor Bahl, director of Microsoft Research’s Mobility and Networking Research, told GeekWire. “Once we get the basic problems solved and get everybody excited, we open up the floodgates … to creating a next-generation traffic management system.” As the software matures, it could get better at identifying factors that increase the risk of collisions, and more accurate at flagging problem junctions.

Microsoft and Bellevue hope to have the system up-and-running by late 2016, with the software hooked into all 200 of the city’s traffic cameras. Based on its predictions, the software might suggest changing the way certain pedestrian crossings are designed, altering the location of traffic signals, or even adding a dedicated bike lane to certain thoroughfares. All those changes will have to be made by people, of course, but the software could take some of the guesswork out of trying to determine what changes will be most effective – and could provide useful data to help city planners justify the expenditures needed to make streets safer.

Sad to say, but similar studies and projects are years away in South Africa if our bumbling road officials even allow something similar to take place.