A self-driving car killed a pedestrian last week in Tempe, Ariz. when an Uber SUV struck Elaine Herzberg as she was crossing the street with her bike at night. Since the accident, students and faculty involved with autonomous transport research at the University of Michigan have grappled with the implications of the fatality and how the event might alter the trajectory or research of driverless cars in the future.
The University opened its own autonomous vehicle testing facility, Mcity, in 2015, a 32-acre urban environment on North Campus with freeways, roads, signs and artificial storefronts. Since then, the international research community has regarded Ann Arbor as a leading city for self-driving research and testing.
Already, Uber has temporarily grounded public autonomous testing in four cities across the U.S. since the fatality yet current state and federal legislation does little to discuss the ethics of publicly testing autonomous vehicles. In December 2013, Gov. Rick Snyder signed a bill into law that advanced the public testing of self-driving cars on Michigan roads. At the time, Snyder was one of several governors nationwide who promoted the testing of driverless vehicles in their states.
Some feel as though the event could bring the testing efforts of Uber and other tech firms to a halt in Tempe and in other U.S. cities that support the technology. On campus, University students have expressed surprise in response to the fatality, but worry little will be done to prevent autonomous vehicle tragedies down the road.
Engineering senior Charu Dwivedi said he thinks the accident will be regarded as a critical moment in the development of the driverless cars.
“Especially with the speed and recklessness that Uber and other companies have been trying to push the technology, I wasn’t very surprised,” Dwivedi said. “I’d say it’s an important chapter that had to happen in the progression of autonomous vehicle technology.”
Information senior Ethan Jannott, who has focused much of his undergraduate career investigating and designing autonomous vehicle interfaces, said he thinks the crash in Arizona was largely unpreventable. Though he was surprised the driverless Uber failed to detect the pedestrian crossing the street, he said it would have been difficult for a human driver to respond in time as well.
“If it was a human in the car, it would have struck her in the same way,” Jannott said. “Because it was really dark, a person wouldn’t have picked up on the pedestrian either. That is only the first time an autonomous car has killed a pedestrian, so it was bound to happen at some point … I think people may be slightly wearier to accept the technology in the near future, but it’s not going to affect the industry.”
Tech companies and automakers like Uber, Lyft, Google, Waymo, General Motors Corp. and Toyota have invested billions into autonomous vehicle research acting on a hunch that soon driverless car will be the norm and humans behind the wheel will be a thing of the past. Politicians, like Arizona Gov. Doug Ducey, have welcomed public autonomous vehicle testing with “open arms and wide open roads” even as much remains unknown about the technology.
Jannott commented while it may seem dangerous to have legislation that encourages public testing of autonomous cars, it’s too important to reject.
“I think public testing has to be done at some point,” Jannott said. “I think we’re right on the cusp of where it’s totally accepted technology. At the same time, the level of autonomy that they’re at is at a point where we can trust it. I think there should definitely be more adoption of the technology around the country for safety reasons.”
Many members of the University’s research community have been anticipating a tragedy like the one in Tempe for quite some time. Lionel Robert, associate professor in the School of Information, who presides over a University study investigating interactions between autonomous vehicles and pedestrians, said the accident was something which he and his colleagues have been preparing for.
“The industry has been holding its breath for a while hoping this kind of thing doesn’t occur,” Robert said. “I would say, they’ve in some ways been getting ready for when this does occur. When someone is killed at the hands of an autonomous vehicle the real question is, what’s next?”
According to Robert, the University holds a unique position in the discussion of self-driving technology because of the resources from different schools that are connected with Mcity.
“From the technical standpoint, we can test the technology in a lot of different areas,” Robert said. “The larger picture is that we occupy a unique position in our ability to engage the public in this conversation. I think Michigan across the board, we have a lot of the ability to lead proactive change in this field because of the strength and diversity of our approach to autonomous driving. The School of Engineering, the School of Information, the School of Public Policy are all getting involved and chiming in on this debate.”
Anuj Pradhan, assistant research scientist at the University’s Transportation Research Institute, is involved in investigating pedestrian trust in autonomous vehicles. He said while the fatality was tragic, it will motivate researchers to perfect future iterations of the technology. He noted that without failure and accidents, those developing the autonomous cars would struggle to polish their product for consumers.
“When there’s a fatality, it means that we are even more motivated to do our research well so that fatalities don’t occur in the future,” Pradhan said. “Our research explicitly focuses on preventing these things from happening so when a tragedy like this occurs, obviously we are very distressed, but overall for safety researchers, we just get all the more motivated to make sure that something like this will never happen again.”