From the April 2020 issue of Car and Driver .
A few years ago, death was knocking on our driver's-side doors. Driving was over. Done. Kaput. We would become passengers in our own vehicles as the autonomous revolution took over our roadways and garages. The message wasn't coming just from Silicon Valley disrupters. In 2016, Mark Fields, then the CEO of Ford, told assembled press in Palo Alto, California, that his company intended to have robotaxis on the road in 2021. Everything was changing. But then it didn't.
The advent of the autonomous vehicle has been delayed more than a few times by more than a few companies. Humans take the ability to manage the cognitive load of driving for granted, but building a computer system that can match our abilities is extremely difficult. It's also incredibly expensive. Years of research and development are still needed before Level 4 autonomy---in which the car can safely perform all driving tasks but only in limited areas---is accessible to consumers. Rushing the technology could have catastrophic consequences.
"The industry in general was much more optimistic, I would say, a couple years back. We've come to the realization that this is not an easy problem to solve." ---Sharath Reddy, Senior Vice President at Magna
In 2018, in what now looks like an inflection point in the development of autonomous vehicles, a self-driving Uber test vehicle struck and killed a pedestrian in Tempe, Arizona. In the wake of that tragedy and with the realization that no amount of hype would solve the technical challenges of autonomy, companies have adjusted their self-driving timelines. After almost two years of collaboration, automotive supplier Magna recently broke off its partnership with Lyft to co-develop an autonomous vehicle. The company will focus instead on Level 2 driver-assistance systems like the ones already found in many new vehicles (in which the vehicle can steer, accelerate, and brake with human oversight). Those can be enhanced to better serve drivers and profit margins.
Manufacturers are backing away from aggressive promises, too. Nissan has admitted that it's unlikely to produce self-driving cars before the end of the decade. And GM, which had previously promised to sell an autonomous vehicle by 2019, has not announced an updated projection. Instead of focusing on fully automated driving, Toyota has, since 2017, relied on a two-track strategy to crack autonomy. One track, called Guardian, is designed to prevent accidents using technology such as automated emergency braking. The other, Chauffeur, is a self-driving research program. Guardian research and development feeds into the Chauffeur program. When Toyota does figure out self-driving, it will have done so via the Level 2 features we're already using.
One company that hasn't deviated from its timeline is autonomous-tech supplier Mobileye. Jack Weast, Mobileye's vice president of autonomous-vehicle standards, says the company is still on track to launch a Level 4 ride-hailing service in 2022.
"The pullback in road testing across the industry is not what will make the technology better. Those who get back on the road and test will take the lead." --- Jeff Schneider, Former Engineering Lead at Uber ATG
Mobileye is betting that the knowledge it has gained from building driver-assistance systems for BMW, Honda, Nissan, and Volkswagen as well as the high-definition maps it has built using anonymized data from vehicles equipped with those systems will help make that goal a reality. Mobileye is also working to prove what it calls "true redundancy," where two test vehicles in Tel Aviv---one that uses only cameras and another that uses only lidar and radar---explore the area and find obstacles. Mobileye can then address any shortcomings of the camera-only car independent of the sensor-based car, and vice versa. When all the hardware is brought together in a single vehicle, the company can be confident that the car can continue to operate even if one of the sensors fails.
Uber is still planning for an automated future, too, though its definition of autonomy may differ from the common understanding. An Uber representative recently told CNBC that the company could consider vehicles that can operate without human intervention on predetermined right-turn-only routes as part of its self-driven fleet.
But overall, the rush to be first has apparently been replaced with business plans adjusted for profit and long-term development and testing. Weast says the industry needs to come together to create common standards for autonomous-vehicle systems before the technology makes its way into consumer vehicles. According to Gill Pratt, CEO of the Toyota Research Institute, there's something even more important than that. "The question I have been asking since I first came to Toyota four years ago is 'how safe is safe enough?' More people in the automated-driving field are starting to grapple with this."
Drive.land is a unique app devoted to cars and their owners. Here you can buy or sell a car, read reviews, watch funny dash cam videos and even find life hacks about car care. We also have a feed with news about cars and new model previews.