Making public transportation faster through algorithms

Fuels |  2 min. read

25 million kids ride a school bus every day, which means there might be about 50 million parents who have an opinion about the crazy route their kid’s bus takes.

Why does the bus pick up kids who live farther from school, after it picks up my daughter?  Why don’t they drop off my son earlier in the route?  Why does the bus go left there?  Or right here?  And who makes up these routes anyhow?

Well, to that last question at least, there’s a new answer and it looks like it’s a better answer:  an algorithm.

The school system is Boston’s, and here was their starting point:

  • more than 25,000 kids get on a school bus in Boston every day
  • they go to 220 different schools
  • the schools start class at times ranging from 7:15 to 9:30 a.m.
  • about 5,000 of those kids are special needs kids, who need to be picked up and dropped off at home
  • you have to factor in Boston traffic, and that’s still not all the variables. But we’ll stop here.

Every year, it took a team of school district staff ONE MONTH to figure out the next year’s routes.  There were that many different possible combinations.

So the school district threw out the challenge to all those supersmart college students and professors and research institutes in Boston. Help us!  Design an algorithm to sort all those possibilities and turn that into the best possible set of bus routes.  And a couple of MIT graduate students did it.

The results?  Boston saved enough money to put an extra $5 MILLION back into classrooms.  It needed fewer buses.  The buses drove fewer miles. Almost one million fewer miles, which meant less wear and tear on the equipment, more efficient use of fuel, and fewer emissions.

And, once the algorithm was worked out it took THIRTY MINUTES to plan the bus routes, which then got a look over from human staff too.

And since it’s Boston, maybe we can say that’s using the old bean.  And if you don’t know the connection between Boston and beans, here you go.