Despite hesitancy with artificial intelligence, Denver Public Schools learned with partner HopSkipDrive to optimize its school-bus routes with AI-driven software.
“I think that there’s a nervousness that AI will take over the world and get rid of the need of routing,” said Tyler Maybee, director of operations for transportation services at Denver Public Schools. “I honestly think this is just seen as a time saver and a way to create a really functional base level of understanding for a particular routing team. I think it’s meant to be a support tool. It’s never meant to replace … I’m not looking to reduce my staff because I use RouteWise AI.”
Maybee explained that during the pandemic, DPS formed a partnership with HopSkipDrive’s alternative transportation services for McKinney-Vento Homeless Assistance Act transportation. He explained that busing students protected by the federal law is sometimes inefficient because they are often located far from their home schools.
It was around early 2022 that the company brought up its idea of an AI solution to DPS.
“We’re a very innovative district, and we always jump on board any type of chance to help push the boundaries of transportation,” said Maybee, who has a data background. Before joining DPS, he served as a data analyst for a non-profit, One Earth Future Foundation.
He explained that when HopSkipDrive presented the RouteWise idea, he thought it could work well for the district.
“We partnered with them, signed an agreement to share data, and we worked with them on parts of the tool that would be beneficial for our business. We created basically the whole thing together,” he added.
Corey McMahon, chief product and technology officer at HopSkipDrive, said RouteWise AI officially launched in 2023 and the software “is a planning tool that helps districts make smarter and faster transportation decisions.
“What it does is it analyzes all their data, ridership, vehicles, traffic policies and it gives them customized routing scenarios that can save them time, money, reduce emissions, reduce student time in transit, whatever their district priorities are,” he continued.
He explained that when districts sign on, they’re paired with the RouteWise AI team of experts, consisting of former directors, routers and consultants from the industry. McMahon said the experts sit down with the district to understand constraints, policies and goals. From there, they build out a custom AI model specifically to represent their transportation system.
“That AI model generates thousands of scenarios, and from those produces optimized results based on their priorities,” he said, adding that districts can then adjust, tweak or tune them through their existing tools. The software is routing software agnostic and integrates with the district’s existing routing software.
A new function is the insights dashboard, which compares data points on a national level. It launched at the start of the year and McMahon said it serves as a spot for district leaders to see all their data in one place. Recently, a national benchmark function was added, which gives district leaders a chance to compare student ride times, cost per student, and bus utilization rates against their peers.
Maybee explained that one of the data points DPS looks at on the dashboard is the utilization percentage. “For us, that’s kind of an important thing, making sure that our buses are full and that our buses are being well utilized,” he said.
Another function is its new feature for on-time arrival. “It kind of gives us the overall picture of which segments (segment is one run of service for a school) in our in a route package that are a potential risk for running late, based on the number of students that are on there … depending on the variability of traffic.”
McMahon said the lateness factor is determined by looking at how tight the routes are. “If everything has to go perfectly for a bus to be on time, it has a high lateness risk,” he explained. “If there is flexibility in the system for unexpected events, student comes out late, late traffic, etc., then your lateness risk is lower.”
By implementing the software, DPS has saved more than $500,000 annually. Maybee said the software is primarily being used to evaluate start times and adjust bell times. He explained that all secondary schools recently moved to an 8:20 a.m. or later start time, and DPS used the tool to inform decisions around bell time adjudgments, which resulted in the savings.
Related: Report Highlights Shift in Federal Policy from EVs to Conventional Fuels
Related: C-V2X Technology Promises School Bus Time, Cost Savings
Related: Canada Becomes First Country to Mandate External School Bus Surveillance Feeds
“We’ve worked with them a couple more times to look at our current state route packages, and they’ve given us a couple suggestions on merging some routes together, or putting bus packages together into one bus and taking a bus off the road that has saved us money,” he said, adding that the software will also recommend taking one student off a particular route to save time. “Their tool is pretty service agnostic. It’s not going to favor putting students on a vendor over putting them on a yellow bus. Their philosophy is generally, if we make a school district’s transportation system healthy, then there’s a natural spot for [alternative transportation] to exist, because there’s always going to be students that are inefficient in the overall system.”
Maybee added that the software is not meant to replace an existing routing provider or routers.
“We’re never going to lose routers because of this software,” he said. “It’s meant to be an additional set of eyes and a starting point to build routes. There are always going to be things that our routers know that that system does not. I have kids that are better suited to be on a yellow bus, and we put them on a yellow bus, even if it’s quote, unquote not efficient, but it’s the right thing to do for that student’s needs. And that’s something that our routers know and that our computer system is never going to know, but it at least gives you an 85 percent solution to get you in the right direction, to provide you the space that you need.”
Data Driven
As for the safety element of using RouteWise AI, Maybee related back to the lateness factor of the program. “In terms of buses running on time, you’re reducing your risk of danger zone infractions, as students aren’t running to get on the bus quickly or crossing a dangerous street.”
He said the software understands the factors that a router inputs, such as right-side only picks up that eliminate the need for students to cross the street, no highway travel, etc.
“It manipulates routes with the data that you provide, so it wouldn’t put a bus stop, for example, in the middle of a busy street,” he said.
McMahon said the challenge for increased safety is the entire ecosystem from the strain of the school bus driver shortage to inefficient routes. “You have a lot of chaos at pickup and drop off, where you have late routes, you have overcrowded busses or underutilized busses, etc. And so, reducing that chaos, minimizing ride times, and making sure that people are where they need to be, when they need to be there,” he said of elements that make pupil transportation safer. “If you have busses arriving late, you have students waiting in unsafe areas. And there’s a number of safety implications, as you can imagine there.”
Maybee added that by using the software, the amount of time it has taken for the routing team to get the entire school year up and running has been significantly reduced. “We kind of have a starting spot with the solution presented for RouteWise AI, that we can then implement as our base layer and build from there,” he said. “We have massive time reductions within our routing team, which allows them to pay attention to detail better, and allows them to take care of the accommodations and take care of the students that need to be taken care of at the highest level, really providing better equity in terms of our service.”
McMahon said one of the greatest elements of the insight dashboard is the availability of more data. “What we’re really trying to do is build a transportation intelligence platform that allows leaders to look at data from all of these sources as they deploy new technology, whether that’s cameras or telematics or ridership tracking or anything out there, they need a way to make sense of it all, and so that’s really where the dashboard is trying to fit in.”
Maybee concluded by saying there is a fear about AI in student transportation. But he cautioned other districts to not feed into the misconception.
“We’re the content experts, and a computer is never going to fully know what we know,” he said. “And I agree with that statement, but I also think that a computer has a better sense of being able to understand the entire picture and giving a solution that has at least the context of the entire picture that you can then kind of manipulate from there. We were surprised with some of the combinations that it came up with. …It never makes sense to just turn on RouteWise AI have it create things and then just like, spit it out and say, okay ‘Driver, go do this.’ That’s not going to lead to beneficial day-to-day service.”
