By Adam Minter, Bloomberg
On a recent Friday evening, a white Toyota Sienna minivan with a cylindrical sensor mounted on its roof slowed to a stop in front of the only hospital in Grand Rapids, Minnesota, population 11,000. The door opened, and I took a seat behind the driver: a computer rack mounted in place of the passenger seat. Next to it was a friendly young operator who sits behind the steering wheel and ensures that this self-driving rideshare doesn’t suddenly skid into a snowbank or a pedestrian. Then we were on the way, passengers in the first autonomous vehicle pilot to run in a cold and icy rural environment.
It’s a pressing need. There are more than 1 million car-less households in rural America. Providing affordable transit to them has always been an expensive challenge. Thanks to rapidly aging rural demographics, it’s becoming harder. Drivers are scarce, costs are high, and the demand for rides to the doctor, the supermarket and the community center is booming.
Rural Americans aren’t the most obvious early adopters for robo-taxis. But right now they need transit innovations far more than people in more densely populated communities, and are far more willing to accept them. For autonomous technology companies, that’s an opportunity to establish the reliability and usefulness of technologies that have struggled to gain acceptance in cities and suburbs. In Grand Rapids, one of those companies, May Mobility Inc., is partnering with government and the community to make that market real. If they succeed, self-driving technologies will have earned a powerful business case, and millions of rural Americans will have a ride.
The Demographics Problem
Located 180 miles north of Minneapolis, Grand Rapids is the biggest town in sparsely populated Itasca County. It may seem counterintuitive that anyone would try to live without a car in such a vast and cold region where health care, jobs and other resources are concentrated in a single town. But the reasons aligned against personal car ownership in Grand Rapids, and across rural America, are powerful.
In 2021, 20% of the 46 million rural Americans were over the age of 65, compared to 16% of Americans in urban areas. Those rural Americans were, on average, poorer than their counterparts in urban areas — and more likely to be disabled. However, even rural seniors who can afford a car and are physically able to drive one are disinclined to get behind the wheel as they grow older.
That creates a dilemma. Car-less or not, seniors and disabled rural residents still have places to go. Non-emergency medical appointments and grocery shopping are critical to maintaining health and independent living. Community-oriented activities, from churchgoing to family get togethers, boost quality of life and reduce pressure on scarce caregivers.
In cities and suburbs, public transit buses can meet some of these needs. But due to their low population density, rural areas are more difficult and expensive to serve well, especially in the evenings and during weekends. For example, the last bus departs Grand Rapids’ only hospital at 3:20 p.m.; anyone with a late afternoon or evening appointment must rely on expensive non-emergency medical transport or a taxi to go home. That intermittent service typically hits those least able to afford it: In the US, 87% of the least revenue-efficient (defined as revenue per passenger mile) bus services are located in rural communities. Of those, 80% are located in communities with median incomes below the poverty line.
In 2019, Minnesota Governor Tim Walz appointed a council to study and advise on challenges related to new transportation technologies, including autonomous vehicles. Myrna Peterson, a quadriplegic disability advocate from Grand Rapids was one of the first appointments. “A while back I started asking why people weren’t at things like community events,” she told me at a Grand Rapids community center she reached via the city’s autonomous shuttle service. “No transport, especially in the evening and weekends. That’s something we need to be independent.”
Around this time, May Mobility, an autonomous shuttle company based in Ann Arbor, Michigan, was looking for a rural community “where we could really demonstrate that we could help,” explains Edwin Olson, May’s chief executive officer, in a phone call. The help, as Olson views it, comes down to replacing or supplementing low-performing buses with May’s on-demand, point-to-point, autonomous shuttles. Olson tells me the cost of May’s shuttles are on par with typically inefficient rural bus services, while providing better service hours and lower wait and trip times.
Much, but not all of the time, that service will be autonomous. GoMarti’s Siennas are equipped with technology (Level 4 automation, in industry parlance) that enables them to drive in most conditions without a human taking over. However, for safety purposes, a human operator remains behind the wheel — mostly observing, not unlike an airline pilot on a highly automated passenger jet — in case conditions, such as iced-over roads, poor visibility, or a roundabout, require it. Over time, performance should improve and the role of the human operator will become less relevant. But even if the vehicles reach a point where they can operate in a white-out blizzard, it’s likely that an operator will remain present to help elderly and disabled passengers access the vehicles. For example, automated securement of wheelchairs remains an extremely difficult technical problem that’s unlikely to be solved soon. For May, the cost of the operator, now and for the foreseeable future, is figured into the model, at least in Grand Rapids.
Minnesota’s Autonomous Rural Transit Initiative (goMARTI), an 18-month, roughly $3.6 million demonstration (half funded by the state of Minnesota with the rest coming from public and private sponsors) began running in September in Grand Rapids. The service offers five specially outfitted Toyota Siennas, three of which are wheelchair accessible and compliant with the Americans with Disabilities Act. The shuttles are free, and can be requested using an app or by calling a dispatch center.
On a recent evening I took several goMARTI rides around Grand Rapids, getting a look at the town and the service. It was a seamless and often dull experience. I watched the shuttle change lanes, turn, stop at stop signs and even negotiate busy intersections. It really didn’t feel much different from being a passenger in a regular car.
For May Mobility, achieving those uneventful rides has been far more challenging. One example: Autonomous vehicles that operate in cities often rely on tall buildings as navigational aids. In a rural setting there are fewer such landmarks. So May Mobility erected what CEO Olson called “totem poles” — simple visual markers — along featureless stretches of the goMARTI service area.
Then there’s the Minnesota weather. On particularly cold days, tailpipe exhaust can look like “mobile obstacles” to autonomous vehicle sensors. Snow and ice present more obvious challenges. Human operators take over when roads are coated. Yet even when the roads are clear, the vehicles struggle with other ubiquitous elements of winter. During one of my evening rides, a shuttle began to veer into a snowy shoulder, possibly confused by the road’s boundaries. Later that same evening, a shuttle dropped me off in a snowbank where, in warmer weather, a sidewalk would be.
When shuttle operators encounter incidents like these, they hit a button on the console to log a record for review by programmers and engineers who will seek to make improvements. Olson calls Grand Rapids “the crucible” where the company will learn to handle snow and ice. So far, it’s learning, and even improving, on human performance in some crucial areas. Two operators told me that shuttle sensors have detected deer about to jump into the road — a persistent danger on Minnesota roads — before they could.
Measures of Success
Ultimately, goMARTI can be judged a success if the people of Grand Rapids feel comfortable and safe choosing to use it. The early returns are promising. According to May Mobility, the shuttles have served 687 people (in a town of 11,000), more than 75% of whom are repeat riders. Equally important, roughly 30% of all rides have included a wheelchair.
Longer term, questions about affordability will inevitably challenge whether such a program is worthwhile. GoMARTI is a free service, but transit subsidies are not unusual in rural or urban areas (New York City’s subway couldn’t operate without them). If, as May Mobility claims, the cost of providing autonomous services is competitive with the most inefficient transit services already offered in rural regions, the upgrade — even with an operator — is worthwhile. Minnesota and Grand Rapids aren’t the only places thinking this way. In Japan, the government and automakers have long viewed the country’s rapidly aging countryside as an important destination for autonomous vehicles; in France, a consortium of companies is preparing an autonomous shuttle program designed to revive its rural regions.
Meanwhile, in the US, the federal government and several universities have been examining rural autonomous transport for years. GoMARTI’s success or failure won’t make or break any of those programs and pilots. But with each ride, it’s building the case for networks of autonomous vehicles serving residents of rural communities, in the US and beyond.