Minnesota will test its autonomous shuttle on icy roads
December 13, 2017
A new pilot program will test the emerging technology in ice and snow.
Tyler Folsom's three-wheeled dream could change how people think about getting around downtown.
Colin Wood is the managing editor of StateScoop. Before that, he was a staff writer for Government Technology magazine. Before that, he taught Engl...
In a lab tucked away somewhere on a college campus, a rangy man with broad palms and wispy blond hair stands over two young men. One of the men comments that the small electronic component he holds in his hand is not functioning properly. There is a bug somewhere. The room's countertops, lined with circuit boards, plastic trays filled with wires and solder, monitors, and boxes trimmed with dials and gauges, are where the builders of the Elcano Project work. Near the door sits the project's centerpiece — two recumbent tricycles, exploding with cables and wires.
The man with the blond hair is Tyler Folsom, an affiliate computer science professor at the University of Washington Bothell who started thinking on the project almost 10 years ago. Folsom participated in the DARPA Grand Challenge in 2005 and 2007, back when autonomous vehicles were more like oversized hobby toys than the branded products that will soon be stamped out to fill the nation's roadways. The size of those machines scared Folsom a little, he admitted, and so he turned his work toward something smaller.
Folsom's autonomous tricycles weigh less than the people who ride them.
But Americans love their big, heavy cars. And as autonomous vehicles pushed by companies like Tesla and Google inch closer to reality, Folsom says government is going about the technology's arrival all wrong.
Today's urban transit systems are not designed for the size of the populations they serve. Seattle, the nearest major city to Folsom's university, exemplifies the necessity of AVs. The streets are packed. The Washington State Department of Transportation reported a 35 percent increase in traffic between 2013 and 2015. There's no space to build new roads in the city and a local technology presence led by Amazon and Microsoft is pulling new workers into the region fast. A report from the Puget Sound Regional Council estimated that 236 people move to Seattle each day. Now passing 4 million residents, the Puget Sound region is in the middle of its biggest population gain of the century and operating on a transportation system that was outdated even before the recent population boom even started.
Similar growth can be seen in urban areas around the U.S. Seattle's treatment of the traffic problem is typical, which is to say the city is trying to come up with solutions, but no one has come close to solving the problem yet.
Mid-2015, Seattle held a hackathon called Hack the Commute, in which the city invited the public to help government come up with inventive solutions. Such events aren't meant to solve local transportation issues outright, but to begin relationships and create new avenues for change that build up ideas to chip away at the problem. But with each day that passes comes another 238 people — more plaque clinging to the city's clogged arteries.
Seattle recently approved a $54 billion project that will bring 64 miles of light rail to the city in the next 20 years, but that project is only expected to serve 30,000 additional riders. Even if the project were completed tomorrow, it wouldn't be enough. At the current pace, America's cities are playing a catch-up game they can never win.
It's not just one city's treatment of the transit issues that Folsom objects to, but rather the posture that government has chosen to adopt. As with many new technologies, AV policy and adoption is contingent on developments in the private sector. Preliminary talks on how to regulate AV have been happening in states for years, but government's habit has primarily been to wait for the technology to arrive and then simply react to the private sector's offers by making regulations.
Leaning back in an office chair next to a slab of books and papers, Folsom argued this approach is stuck in old ways of thinking that limit the technology's potential.
"I'm trying to flip transportation. I'm seeing that cars are going to become automated and once we have automated cars as a fact of life, what the car is is going to change," Folsom said. "And this very light vehicle might be the vehicle of the future. If you look at horseless carriages, the very first horseless carriages looked like carriages, not at all like modern cars. I think once we get to automation, we're going to change what a car looks."
Folsom's idea is to install fleets of light-weight autonomous vehicles where today's bus lanes, bike lanes and high-occupancy vehicle (HOV) lanes are, and operate them similarly to public rapid transit (PRT) systems found in Europe and on some private campuses in the U.S. Folsom approaches the problem like any good computer scientist would, putting the problem front and center and then using what tools are available to meet the goals outlined. By his calculations, every aspect of light-weight AV is superior to traditional urban transit — from safety, to cleanliness, to energy consumption, speed, cost and range.
For starters, creating dedicated lanes for light-weight AVs greatly diminishes the risk of being crushed by a 5,785 lb. Cadillac Escalade. Using three wheels instead of four was an "attractive" way to keep the price and the weight down, Folsom explained. Two- and three-wheeled vehicles are also subject to fewer regulations. Building on four wheels, Folsom may have found himself in discussions with the Department of Motor Vehicles (DMV).
And improving on the speed of urban transit should be easy. Americans spent 6.9 billion hours stuck in rush-hour traffic in 2014, according to a 2015 report from the Texas Transportation Institute at Texas A&M University. And the Environmental Protection Agency makes its urban transit calculations with an assumption of 19 mph. Folsom predicted that a dedicated urban AV lane could move at a constant 30 mph.
Lighter vehicles are also more efficient. The ratio of fuel used to move vehicles rather than the passengers themselves is vastly tipped toward the vehicle side. Folsom calculated that a light-weight AV vehicle modeled after his prototype could operate with one-tenth the energy consumption of today's road vehicles.
Deaths on America's roadways have declined in recent years, but motorcycle deaths have increased. Almost 5,000 people died in motorcycle crashes in 2015, up 8 percent from the previous year, according to a National Highway Traffic Safety Administration report. Motorcyclists were also found to be 27 times more likely than passenger car occupants to die in a crash per mile traveled. Most pedestrian deaths — there were more than 4,700 in 2013 — also occur in urban areas, according to the Centers for Disease Control and Prevention. Light-weight AV could ameliorate these issues too, Folsom said.
"If you cut down the traffic accidents that you have, it suddenly becomes safe to ride a motorcycle, because the [autonomous] cars are going to see the motorcycle even if the driver doesn't," he said. "And they're not going to hit it."
Folsom's eccentric appearance is offset by his demeanor when he talks about the possibilities of AV. He's all business. His speech is measured, almost solemn, and he uses frequent figures and statistics to support his thoughts when questioned on whether something as whimsical as a tricycle could take hold as a common mode of transportation.
"I'm convinced that you can approach the capacity of light-rail with these things, because the capacity of a single lane is typically 2,000 vehicles an hour," he said. "If you automate things, you can double capacities and get 3,000 or 4,000 vehicles an hour. Light rail tends to have capacities like 16,000 riders an hour. You can pack two lanes of these vehicles in the space of one light rail right of way, and if you want you can stack them in four lanes so you can carry as many people as you could on light rail and you can go faster."
The technology behind the Elcano Project holds immense potential, but it can be hard to imagine pitting it against a Tesla Motors valuation that now approaches one trillion dollars. The Elcano Project's current object detection system is powered by sonar, but that's just a placeholder, Folsom said, while they search for a partner — a camera-based system like those developed by MobileEye would be ideal. Current project funding comes from a $75,000 award from the Amazon Catalyst fund.
It takes a vivid imagination to transport those two half-finished bikes from the dusty lab floor to the future Folsom envisions — chains of multi-colored and canopied pods slaloming through Seattle's hillsides, pavement shining under the sun after a brief shower, a single pod occasionally peeling off to its destination as the others continue onward, never slowing down as riders can be seen tapping their smartphones and sipping coffee.
Dylan Katz, a computer science student at UW Bothell, was sitting in front of two computer monitors, listening to Folsom talk to a man who had made an unscheduled visit to offer his assistance on the project. Much of the project's development is done by students like Katz in conjunction with their coursework.
Katz helped code the C++ that allows the vehicle's disparate components and open source Arduino boards to communicate with each other.
"The problem we were having when we first got here is that in order to actually send communication from one device to another device, it would have to do a block, which would basically stall the device while it reads," Katz said. "And under my model, there's no longer a block in place so it's able to communicate more effectively. … It appears to be working pretty well now. What I'm working on now is rewriting the actual codebases that run on the devices so that they actually use it and will allow us to start testing the whole system."
Fortunately, Folsom said, the choice between driving a Tesla and cities building networks of fast-moving tricycle pods is not mutually exclusive. And unlike the billions Seattle is spending to build new light rail, the infrastructure for lightweight AV is ready. The roads are already built. The Elcano Project could be used as a transition between the horseless carriages people drive today and the vehicles of the future.
"You can think of these not as self-driving bikes or tricycles but as a low-risk platform for developing automated vehicles," Folsom explained.
Now operated by a remote control that sends commands to complete routines, like complete a circle, Folsom said he hopes to operate his tricycles fully autonomously by this summer, though they probably won't be ready for real-world road conditions yet.
Seattle is supportive of new ideas in transportation, but as for adoption of Folsom's idea, they're at least a few years off from being ready for something so experimental, though it might work on the city's expanding cycle track network, said Evan Corey, shared mobility program manager at the Seattle Department of Transportation. Seattle cancelled its bikeshare program this year and diverted funding into pedestrian and bike lane improvements.
Folsom's critique of government's reactive role to an emerging private market is supported by recent developments. Companies like General Motors and Tesla are jousting to control the new market as it develops, while the U.S. Department of Transportation and National Highway Traffic Safety Administration released draft guidelines for AV in September that will provide a mere starting point for many state and local regulations.
"This field is pretty nascent," Corey said. "And it's so nascent that there isn't any finalized federal regulations for the industry. One thing in my review of [the draft guidelines] that has been evident is that it is woefully behind in terms of accommodating the needs and context of urban areas. My assumption is that the federal government is going to put a lot of stress on local municipalities to set regulation on things like right-of-way management and intersection control and some of the aspects that relate to design of our streets."
Small-scale autonomous vehicle pilots have been underway around the nation for several years. From Google's early experiments that now include a pilot in Seattle-neighbor Kirkland, Washington, to a hodgepodge of smaller companies carving out their respective niches, like a nuTonomy pilot now underway in Boston, the technology is propelled by America's love of cars.
Seattle is probably a couple years off from a pilot like that, Corey said, citing the region's hills, rain and poor lighting in winter. In transportation, he said, the chief motivator of new technologies is safety, but AV isn't the end game.
"In an area where you constrain land mass, you have limited right of way, you're still going to see traffic, even with autonomous vehicles, and traffic will continue to be exacerbated as growth continues in the region," he said. "There's no way around it. The need for transit is going to continue to be evident in the Seattle region regardless of if we have or do not have autonomous vehicles operating in our streets."
Beyond safety, beyond technical barriers, funding, policy or public approval, the Elcano Project's biggest challenge is cultural. Americans love their cars and they love autonomy. But Folsom had an answer to that, too.
"If you take this to the next level," he said, "you can basically set this up so the vehicles operate automatically on the line they're designed for, and then when you get to the station, take them off and drive them manually to where you want to go."