There's more and more grumbling out in auto land that car and tech companies are taking several cautious steps back from the autonomous vehicle precipice.
Last summer, tier one technology companies (those delivering many of the subsystems and perception technology to automakers) were trying to reconcile themselves to the idea that AI (artificial intelligence--really just machine learning) was not going to solve the problem of so-called edge cases, those unique situations we encounter on the road that don't follow the rules but which can be fatal if not properly navigated. (Picture a mattress falling off a truck in front of you on the highway.)
One of the busiest weeks in ride-and-drive programs last week had everyone from Audi to Honda on the road (see the new 2019 Honda Passport review here this week) showing off their latest models on the road and off. And again, the apres driving discussions included muted conversations about how overly optimistic automakers have been about delivering autonomous vehicles to consumers. That 2020-2021 time frame, many were saying, was too sanguine by far.
Adding to the more cautious consensus, Apple revealed it was diverting 200 employees away from its self-driving software development efforts late last week. That's not significant news given how many other companies have more experience and have been working on the technology with greater resources than Apple, but it is a sign that companies that think they can just parachute into the autonomous space with money and processing power are in for a rude awakening.
So what's the new timetable look like? Much right now depends on how many real-world miles (versus simulated miles) testers can put under their belts and how accurate they can make the vehicles using primarily rules-based systems that resort to machine learning algorithms as little as possible. That may take more time than many anticipated.