The greater Ann Arbor area is increasingly well-known as a mobility industry hub, but the actual number of local mobility companies is startling. Officials at Ann Arbor SPARK estimate there are at least 100 mobility startups in the area, ranging from brand-new ones to more established companies.
Included in that 100 are companies who do not serve the mobility industry exclusively but whose technology can be used in autonomous vehicles or other mobility applications, as well as companies headquartered elsewhere that have a strong presence in Washtenaw County.
The growing ecosystem of mobility companies prompted SPARK to hire Komal Doshi as its manager of mobility programs in November 2017. Doshi says numerous established local mobility resources are part of what attracts mobility startups here.
"We have a lot of assets in the region already, ranging from vehicle testing environments like MCity and (the American Center for Mobility) to cybersecurity (firms) that create automotive control systems and auto security, as well as an amazing amount of research and development centers," she says. "That cluster of activity makes it an interesting hub."
We talked to five Ann Arbor mobility startups about the technology they're working on and the trends they're observing in the local mobility ecosystem.
The "last mile" can be a major sticking point in many people's commutes, as they try to find a convenient way to make it from a bus stop or parking lot to their final destination. May Mobility is out to solve that problem with self-driving microtransit fleets traveling at speeds up to 25 mph.
"These automotive companies are realizing it's a long way out before we'll be able to have full autonomy coverage, but there are a lot of things autonomous technology can do today," says May Mobility COO Alisyn Malek.
The company's technology has been demonstrated in Detroit; Tampa; and Frisco, Texas. Test scenarios have included transporting employees from a parking garage to their office and back, as well as moving people around within a large city.
Malek says May Mobility is based in Ann Arbor in part to take advantage of computer science talent coming out of the University of Michigan as well as mobility resources like MCity and SPARK.
She's not worried about competition, despite the density of mobility startups in the region.
"We can learn a lot from each other," she says."For instance, bike share companies are not competition to me, but if we know of a community that could use each other's services we can share that information and make a much stronger network."
Malek is concerned about the lack of early-stage financing in the Midwest as a whole. Some of May Mobility's first investments came from Detroit, but much more of the money came from out of state. She says May Mobility is in a place where grant competitions and other smaller funding sources just can't raise the amounts the company needs.
"There are still a lot of opportunities for the ecosystem to grow in how it supports companies that are looking to scale," she says.
Not every local mobility company is working with autonomous vehicles. Movatic has found success in streamlining the way people share and park their bikes.
Founder Ansgar Strother's previous project was A2B BikeShare. While in the process of selling that company in 2016, Strother says he frequently heard from others in the industry that there was a need for good software to support bike parking.
"If I'm on a corporate campus, why am I downloading one app for bike parking, one for bike sharing, one for a shuttle, and maybe even a fourth app for the parking deck?" he asks. "If you're an administrator, it doesn't make sense to be managing that from a lot of disjointed platforms."
Movatic provides one software platform that helps make bikes and racks shareable and available on demand. Since Strother began running Movatic full-time in 2017, more than 1,300 bikes or bike parking spots have launched on the platform.
Strother says the Movatic platform has many other potential applications beyond bike share programs, and the company probably will expand into other areas through existing partners. Electric bike share programs are an obvious next step, for instance. Another application would be a "shared mode" program where different modes of transportation – for instance, a shared bike or a shuttle bus – are available to a user through one app.
Movatic had a two-person team six months ago, which has grown to four. With the company's customer base growing rapidly, Strother says he expects to add even more employees in the next year or two.
Derq Inc. is one of multiple national and international mobility companies that have a strong Ann Arbor presence. The Dubai- and Detroit-based company is currently a "virtual tenant" in the SPARK Central Innovation Center and it also maintains a presence at MCity.
Derq's technology uses driver behavior modelling, sensory inputs, and vehicle-to-vehicle or vehicle-to-infrastructure communication to anticipate and prevent accidents. Co-founder and CEO Georges Aoude developed the technology as part of his Ph.D. work at the Massachusetts Institute of Technology.
Derq's technology can piggyback on cities' existing infrastructure, such as sensors or cameras used for security or for counting traffic at intersections. Will Foss, Derq's director of business development and partnerships, gives the example of a pilot program in a major city focused on predicting when someone will run a red light.
"You feed that information into Derq's system, and with our machine learning model running in real time, it can predict two seconds in advance before a car is about to violate a red light," Foss says. "It can send an alert, in real time, to all vehicles equipped to receive, giving them two or three seconds to react and avoid danger."
Foss says hardware that detects those signals will become part of a trend in new vehicles, similar to the way onboard GPS systems can already alert drivers to traffic jams in real time.
Foss sees a trend around regional hubs, noting that people on the west coast of the country don't talk about the specific city they are in but refer to themselves as being in "Silicon Valley."
"It's really helpful for us to think less about Ann Arbor and Detroit but more about southeast Michigan as a community we're part of, and we benefit from that," he says.
Humans can describe a scene from a video perfectly, but the man-hours involved in having a human label a video frame by frame are prohibitively expensive. Voxel51 founder Jason Corso has spent the last 15 years training computers to do that work faster and cheaper.
Voxel51's software can identify scenes and interactions in a video, recognize and track objects in the footage, and then provide descriptions of what it sees in natural language. At that point, a human can do fine labelling on the subset of information needed for a specific project or need.
The technology has multiple applications, including automotive. Corso says the technology isn't made to be used in real time on the dashboard of an autonomous vehicle, but rather calls its potential application "forensic."
"You can take video data from the city, watch it, and even track a specific car moving through town to find its trajectory and analyze patterns," Corso says.
He expects that at some point in the future Voxel51's technology can be used to evaluate the vision systems other companies are putting into autonomous vehicles.
Corso says most in the mobility sector aren't worried about competition and are more interested in collaboration. Voxel51, for instance, plans to release part of its video analytics platform as open-source code in June.
"Building an open ecosystem of innovation lets everyone benefit," he says.
We may still think of driverless cars as vaguely futuristic, but PolySync's DriveKit vehicle control interface can turn your vehicle into a remote-control car with a simple one-hour installation.
PolySync was founded about five years ago in Portland, but it's since added an Ann Arbor office in order to build closer relationships with the auto industry and its original equipment manufacturers (OEMs). OEMs and early-stage tech companies now use DriveKit as a test platform during their R&D operations.
PolySync's other flagship product, the Core data capture platform, allows an engineer to collect inputs from over 33 of the numerous sensors that allow connected and autonomous vehicles to interpret their surroundings.
Another technology under development at PolySync would create "fault-tolerant" software to address the functional safety of autonomous vehicles.
"In an autonomous vehicle, what happens if there is a fault?" asks Jay Ellis, PolySync's director of business development. The fault could be in the hardware or the software, and it might not be clear if the fault means a driver should pull over immediately or if the system is resilient enough to run safely until the fault can be diagnosed.
Like others we talked to in the industry, nobody at PolySync is worried about competition. Ellis thinks that a density of tech companies is a good thing.
"Having support and an ecosystem means you know people who can help you deal with a problem or find more information," Ellis says. "It's helpful to have people in the same position when you're commercializing a unique product to address a unique problem. That ecosystem makes all the difference."
The future of mobility in Ann Arbor
Doshi says she hopes the future of mobility in southeast Michigan will include a regional transit authority and better connections between Detroit and Ann Arbor.
She also thinks tech companies need to not only collaborate with each other, but to create more public-private partnerships.
"There is really a great opportunity at this point for these different partners to work together," she says. "We need both cities and the private sector to be better engaged and start pumping more resources to get quick, dynamic results."
Sarah Rigg is a freelance writer and editor in Ypsilanti Township. You may reach her at firstname.lastname@example.org.
All photos by Doug Coombe.