Imagine pulling up to a stoplight only to notice there’s no driver behind the wheel of the car in the next lane. That could soon become a common sight in Michigan now that Gov. Rick Snyder has signed into law a new measure allowing driverless vehicles to operate on public roads.
Michigan has long been the center of the auto industry, and it hopes to retain that role at a time when the automobile is rapidly becoming what National Highway Traffic Safety Administration chief Mark Rosekind has called a “supercomputer on wheels.” Consider that the average 2017 model will use about 100 million lines of software code, several times more than the latest U.S. Air Force fighter jets. By the end of the decade, those software codes are projected to triple.
A key reason: the arrival of the autonomous vehicle. Long the stuff of science fiction, the robo-car is rapidly becoming a reality. Several automakers, including Tesla and Mercedes-Benz, already offer systems that allow limited hands-free driving. By 2020, Nissan plans to have as many as 10 fully autonomous models on the road, while Ford promises a fully driverless car by 2021.
All of those new entries will revolutionize the transportation industry. Uber, for one, is betting that by taking drivers out of the equation it could bring the cost of ride sharing down so low it would make more sense to use the service than own a car. Delivery services like UPS and FedEx are also studying their options.
But are we moving too fast? That’s a question many observers are beginning to ask — even some of those who believe that, in the long term, autonomous vehicles will sharply reduce the number of deaths and injuries on American roads. Stirring the safety debate is a series of relatively minor accidents involving fully autonomous Google prototypes, as well as a fatal crash of a Tesla Model S running in so-called Autopilot mode last May.
Despite the rush to get fully autonomous cars into production by the start of the coming decade, “We’re a long way from the finish line,” says Gill Pratt, CEO of the Toyota Research Institute in Cambridge, Mass., and Palo Alto, Calif. Pratt sees lots of issues challenging researchers and engineers. Among other things, even the most sophisticated vehicle sensors today find it difficult to tell the difference between sand, snow, and ice. And though all can be slippery, each must be handled quite differently.
It’s been said a camera never blinks. Perhaps, but even the best sensor systems still make mistakes. Preliminary data suggests the camera on the Tesla Model S involved in the May crash confused the white truck it hit with the bright
Florida sky, and its back up radar thought the semi trailer was a highway sign.
But perhaps one of the biggest challenges will be teaching AI systems that will run autonomous vehicles to better understand how human drivers think. That became an obvious issue as part of the autonomous vehicle development program run by tech giant Alphabet Inc.’s Google subsidiary.
Considered by many to be the most advanced self-driving program out there, it nonetheless has been involved in nearly 20 crashes over the last six years. All but one has been blamed on the other driver, at least technically. But the Google prototypes (being developed with the assistance of Roush Industries Inc. in Livonia) may have contributed to some of the crashes by obeying traffic laws in ways human drivers don’t.
At stoplights, for example, the Google vehicles have been taught to stop immediately if the light turns yellow, but in many situations — in heavy traffic in urban regions — motorists will often make a left turn at the last moment.
Most experts believe such hurdles can be cleared eventually. And Michigan is playing a key role in the process. The University of Michigan in Ann Arbor, has set up a major autonomous vehicle testing complex, dubbed MCity, on its campus. In November, a groundbreaking ceremony was held for a large testing center on the site of the old B-24 plant at Willow Run Airport in Ypsilanti Township.
Google, meanwhile, is establishing its own research center in metro Detroit. It’s not a stretch to reason that Google may strike a deal to build autonomous vehicles with Ford, given the automaker is a major customer of Roush.
Still, the process isn’t easy. Apple was rumored to be building an autonomous car, but now it appears it will sell its driverless technology to any number of automakers.
Going forward, Michigan must work hand-in-hand with automakers and suppliers to retain and attract the best talent. Otherwise, we could be looking in the rearview mirror of an industry that once made Michigan great.
Paul A. Eisenstein is an automotive journalist and publisher of TheDetroitBureau.com, located in Pleasant Ridge.