Boston — Tuesday, July 25, 2017 — The Boston Public Schools is proud to announce that a team of analytics and optimization experts from the Massachusetts Institute of Technology has won the first-ever BPS Transportation Challenge by developing a computer-based model that more efficiently routes school buses, generating potentially millions of dollars in cost savings that will be put back into classrooms.
The “Quantum Team” from MIT's Operations Research Center (ORC) was among a field of local and national computer-science and logistics specialists who competed in the Transportation Challenge, the three-month hackathon-style initiative that BPS kicked off in April to generate routing efficiencies within the district’s complex transportation system and to explore adjustments to school day start- and end-times.
MIT’s winning team consisted of Professor Dimitris Bertsimas, co-director of the Operations Research Center, and Arthur Delarue and Sebastien Martin, both PhD students at the center. They captured the $15,000 Round 1 prize by creating an algorithm that greatly minimizes the number of bus routes by strategically reconfiguring bus stops, maximizing the number of students riding each bus, and reducing the amount of time buses travel when no students are on board.
“This is a huge win not only for the MIT team, but also for the City of Boston and our students. This new model of routing buses will free up millions of dollars for reinvestment back into our schools,” said BPS Superintendent Tommy Chang. “It will also reduce traffic and carbon dioxide emissions by decreasing the number of buses needed, helping improve our environment and ease congestion on our roadways.”
Boston has one of the most expensive per-pupil school transportation systems among school districts in the country. The $120 million spent on transportation in fiscal year 2017 reflects an approximately $33 million, or 7.5% annual increase, over 2011. Finding greater efficiencies in the the routing of BPS buses has been a major focus of the district’s long term financial plan, “Investing in Student Success,” that was released in the fall of 2016.
That report found that while assignment policies help determine how many students are transported and how far, another key driver of transportation costs is the level of efficiency in the system, or the number of buses needed and the number of trips each bus makes daily. A more efficient system uses fewer buses for the same number of students, and therefore buses that were sometimes less than half full previously would now be operating at closer to capacity.
Currently, BPS transportation staff manually build school bus routes using what is considered an industry leader in pupil transportation software. This is a process that takes several weeks to complete. The new MIT solution produces the routes in approximately 30 minutes.. However, the data produced by the MIT system will still be fed into the existing software in order to pressure test the new model and ensure reliability.
By implementing the Quantum Team’s solution, BPS could save as much as $5 million through the reduction of 50 or more bus routes, which will help provide more funding for schools. (A bus route is made up of individual trips to and from schools. The district’s goal is to have its buses run trips to three different schools each morning and afternoon.)
Overall, the new routing model is expected to improve the on-time performance rate for BPS buses, getting students to and from school on schedule much more consistently. The distance that students walk to and from their bus stops is not anticipated to change, on average. In some cases, bus stop locations might change slightly for some students, while remaining the same for others. The MIT team took steps to ensure that no bus stops are sited in unsafe traffic locations.
The Quantum Team solution could lead to a 20,000-pound reduction in carbon emissions produced by BPS buses each day, and could remove nearly 1 million miles of traffic-clogging bus trips off the road each year.
“Our solution generates thousands of possible routes, and then picks from trillions of permutations the optimal bus route to connect schools throughout the day. Creating this many permutations by hand simply is not possible. Our algorithm creatively combines optimization theory, human intuition and the power of computing,” explained Professor Bertsimas of the Operations Research Center at MIT.
BPS Chief of Operations John Hanlon said he is very excited about how the computer model designed by the Quantum Team will help transform the way the district routes buses, with little cost to BPS since the prize money was contributed by private funders.
“This is a sign of a new era in government where we work collaboratively with some of the brightest minds in academia and industry to solve challenging problems in a way that helps us reduce costs while also better serving our families and schools,” Hanlon said. “That aside, we are not simply accepting out of hand what the Quantum Team produced. We are carefully vetting the team’s solution with the decades of real-world routing experience of our transportation staff. Their invaluable knowledge of the ins and outs of Boston’s streets is essential in getting us as ready as we can be for the first day of school.”
BPS launched its Transportation Challenge on April 1, during an event at District Hall in Boston’s Innovation District, as a way to harness cutting-edge optimization solutions from some of the nation’s leading experts. BPS challenged participants to create a more efficient routing algorithm from within the current parameters of our system. BPS also asked participants, as part of a second round of the challenge, to come up with an optimal set of bell times for the district’s 125 schools. The Quantum Team also won this phase of the challenge, which similarly carried a $15,000 prize. BPS plans to release the outcome of this work in the fall, through engagement with the broader BPS community.