The need for a car to navigate the Motor City is never surprising. With the sheer size of Detroit, construction, wide roads, and visiting drivers, be they from across the state or across the border, hazards are not unexpected. Those issues can all add up faster for older or disabled residents.
Detroit has taken a new step into the future of public transportation to help those individuals with a shuttle completes with an automated driving system, or ADS. These self-driving shuttles will be implemented into a yearlong pilot program starting this summer.
The shuttles, a cross between a city bus and a rideshare, run along fixed stops, but instead of the usual bus schedule, they can be summoned by app, phone number or website provided to those who take part in the program. The shuttle will alert them how far away they are from the stop, and pick up the rider.
“(The shuttle) can handle changing lanes and making turns,” says Tim Slusser, Detroit’s Chief of Mobility Innovation. “It can do everything a human driver can do on the streets of Detroit.
To participate in the pilot, residents must be 65 or older or have a recognized medical disability, and live in one of the following neighborhoods:
- Islandview
- Eastern Market
- McDougall Hunt/ Greater Villages
- Lafayette Park
- Elmwood Park
- Virginia Park Community
- Russell Woods
- Boston-Edison
- New Center
- North End
- Elijah McCoy
- LaSalle Gardens
- Dexter/Linwood
These neighborhoods were chosen after research found that they were the ones with the highest number of residents that fit the program’s parameters. Eventually, the program hopes to be available across the city as a whole.
When the pilot begins, major stops will include Eastern Market, the Detroit Medical Center, the Dingell Department of Veterans Affairs Medical Center, Rivertown Market, and Grand Circus Park. Others will be folded in as the program moves forward.
If the idea of a self-driving car makes you nervous, you are not alone. Earlier this year a AAA survey found most drivers are afraid of self-driving cars, 66 percent according to their website. Fueled largely by reports of crashes and failures in testing.
Slusser was expected to run into that concern, particularly because of people's relationships with driving their own vehicles in a city like Detroit. However, when he went out into the neighborhoods to discuss the project, he says he found a different response. People were excited at the idea -- and many of those he talked to, especially the elderly, didn’t like using ridesharing services like Uber.
“'I get picked up by drivers I don’t trust. They drive erratically,'” were the most common complaints Slusser heard.
The ADS Shuttle has a human safety operator called an Automated Vehicle Operator (AVO), which may have eased concerns. They can take control of the vehicle if problems occur, but primarily acts as a concierge for the riders, providing them with any service or information they may need.
While the plan is to have every turn and stop done by the car, if any issues occur tapping the brake or turning the wheel will shift the car into manual.
There are a series of requirements AVO applicants must meet before they are fully instated.
First, there is a comprehensive online course, consisting of 15 videos, which includes content from the Society of Automotive Engineers, The National Highway and Traffic Safety Administration, and the National Disability Institute.
Then, there is a 20-hour course, consisting of five on-site training sessions. This includes manual safe-driving certification and complex staged autonomous interactions. Finally, on their first day on the job they cannot receive less than a 4.7/5 on a feedback survey that relates to their preparedness.
The shuttles have been tested in real world conditions by their designers, May Mobility of Ann Arbor. The tests were conducted at the Mcity Test Facility at the University of Michigan, which is the first Automated Vehicle (AV) testing center based around real world conditions.
Tests included a “Driver’s License test, which is like the test we all take at the Secretary of State’s office, and a “Driving Intelligence Test.” The latter puts the vehicle into the scenarios most likely to cause accidents, especially those that cause death and injury, to see how the vehicle’s software reacts.
Potential risks the software faced included an angle crash with a cut-in vehicle, failure to yield when entering a roundabout, a head-on crash at an unprotected left-hand tun intersection, near miss and angle crash when entering a roundabout, angle crash when entering a roundabout, and a rear-end crash at a signalize roundabout.
To keep the test realistic, Mcity keeps the risks coming in random times, to simulate the chaotic nature of driving.
“We randomize the variables to keep (the car) from knowing what we are testing,” says Gary MaGuire, Managing Director of Mcity.
Without the randomization, the system may learn what to do at that point of the test, not when that type of risk occurs.
Test have also been done on Detroit city street tests this year to see how the vehicle reacts driving on the streets among real drivers capable of real human error, not on a controlled track.
Like any rush hour in the Motown, unexpected variables arose in the tests, but the Mcity tests prepared it, and the vehicle responded as intended.
All of this is done before the elderly and disabled of Detroit set foot onboard, and with them specifically in mind. They are being especially cautious when it comes to Mother Nature’s hazards.
“Right now, our operational design domain (ODD) does not allow us to operate in Detroit when there is snow, rain or ice,” Satvir Singh, Director of Product Safety at May Mobility, said in an email. “We will continue to operate the service in manual mode during those conditions until we have updated our ODD at a later date.”
May Mobility’s approach to AV’s differs from most of the industry, with a unique ability to impact communities.
“We are unique from many other AV companies because of our focus on filling gaps in established transportation systems and improving mobility among underserved populations,” Singh says. "This means that instead of a robotaxi model, we work with cities and businesses to create microtransit zones that address the specific needs of the communities where we deploy.”
Over the course of several months field autonomy engineers spent mapping out the streets that were being used and then two solid weeks testing and tweaking the system. No safety issues arose, but the driving data was able to help make a more comfortable ride.
There is a long-term goal to have the system be available to the wider number of Detroit residents and may include a door-to-door feature. However, for now, the pilot is taking it slow. The year-long process will gather as much information as possible to give the elderly and disabled the best system possible.
That is also why they have a minimum goal of 100 participants. to get a large enough data set.
The data collected may go even further than helping the shuttle use in the city. If successful, Slusser believes the collected information could be used to help plan out insurance cases with self-driving cars in the future.
And the program may be finding interest closer to home than they thought. Slusser says after signing some of the necessary legal documents for the program, one of the lawyers, who is disabled, said he did not have to be a part of the program, but hoped to take a ride at some point.
The application form for the pilot can be found here.