Industry, Experts Converge on 2020 Vision
A University of Michigan-Dearborn researcher polled about 500 people over the past 18 months with regard to safety, liability, practicality, human factors, cost, regulation and the level of automated-vehicle technology consumers would accept.
October 22, 2014
DETROIT – Whether he spoke with mobility experts or people working in the auto industry, a University of Michigan-Dearborn researcher found agreement that the era of automated vehicles would begin in 2020.
Steven Underwood, the school’s director or transportation and information, polled about 500 people over the past 18 months with regard to safety, liability, practicality, human factors, cost, regulation and the level of technology consumers would accept.
Many of the respondents were engineers contacted through SAE International, and Underwood also polled 260 people at the Automated Vehicles Symposium hosted in July in San Francisco by the Transportation Research Board.
In addition, Underwood interviewed the world’s top 20 automated-vehicle experts.
“2020 is the date for introduction of all these systems,” Underwood tells a crowd here focused on electronics at SAE 2014 Convergence at Cobo Center.
He defines the initial step toward self-driving vehicles as driver assistance, a suite of technologies (such as collision avoidance and lane-keeping) already appearing in production vehicles that can control brakes and steering. Partial automation requires more advanced control of brakes and steering.
Underwood separates self-driving vehicles into three categories:
Conditional automation – a system that pays close attention to the environment but still needs the driver as a fallback in case of an error. An example is a fleet of ACC-equipped vehicles driving in tandem on the freeway.
High automation – monitors the environment and relies on itself in the event of a system failure and requires less involvement by the driver. A platoon of cargo trucks on the highway could be highly automated.
Full automation – system provides full-time monitoring and fallback, requiring no involvement from the driver, such as a robotic taxi.
His research suggests the first two types will be in the market by 2020, but that the era of fully automated vehicles will not begin until about 2030. In 2025, Underwood says highly automated cars that drive themselves on urban highways and surface streets should be available.
In the near future, a significant step should come by 2018, when highly automated shuttle vehicles are expected to provide limited transportation, such as on college campuses.
“This particular application is low-speed, and the vehicle would not be engaging with other automobile traffic. It’s on a separate campus,” he says.
“It may engage with pedestrians at low speed. These systems, the ones I’m familiar with, are so cautious and defensive they will stop for pretty much anything. So the risks associated with them are pretty minimal.”
Among Underwood’s other findings:
By far, legal liability is the most difficult barrier facing automated vehicles. But the 500 people surveyed did not identify cost, social acceptance, roadway infrastructure or even the enabling technology to be significant hurdles.
Fully 27% of respondents also agreed automated vehicles need to be demonstrated as twice as safe as today’s conventional vehicles before being authorized for public use. The second largest group of respondents (18%) said automated cars must be perfectly safe.
Respondents differed on whether society will accept that automated vehicles occasionally cause accidents and even fatalities. Some 52% of SAE respondents said society would accept that reality, while 73% of participants at the San Francisco symposium agreed with that sentiment.
Between 2013 and 2025, 31% of survey respondents expect the number of fatal accidents to fall more than 50%.
You May Also Like