As any commuter can tell you, everyday driving is often tedious, and being a passenger can be doubly so. Now imagine combining the worst of both worlds: a driver’s need to focus on blisteringly dull traffic while simply being a passenger most of the time.
That mix is the next step in autonomous vehicles, coming to a forecourt near you soon.
Advertisement
The US National Highway Traffic Safety Administration (NHTSA) uses a six-level classification scheme for vehicle autonomy. Level 0 is no automation. Level 1 includes basic driver-assistance systems like adaptive cruise control. Level 2 allows for simple steering autonomy with constant driver oversight (like Autopilot in Teslas).
Level 3 allows the driver to safely turn attention away on some road types, but requires them to be ready to take over again if needed. It’s the next step in the evolution of commercial autonomous cars. Audi is by 2018. Other big car-makers including Nissan, Honda and Kia are set to follow.
However, Ford has cast fresh doubt on the wisdom of this progression. It discovered while testing level 3 systems that drivers lose “situational awareness”, sometimes even falling asleep, if they try to pay attention without having to be in continuous control of the car. That renders them unable to take over swiftly if required. The problem persists even after adding buzzers, vibrating seats, and a second engineer to watch the one in the driver’s seat.
As a result, Ford’s chief technical officer, Raj Nair, has restated the company’s and bypass level 3 completely. By 2021, he wants to make fully autonomous cars lacking pedals and steering wheels.
The intent is to create a vehicle suitable for ride-sharing or ride-hailing, presumably within a limited, predefined area: well-mapped city streets, say. That’s level 4 autonomy. Passengers would be free to chat, play on their phones or nap without ever worrying about needing to take control of the car… because they ’t take control, negating the twilight zone of level 3.
Level skipped
Skipping the possible pitfalls of part-human, part-machine that level 3 implies sounds tempting (and for a while). But equally, as car-makers race to get models out and capture consumer interest, it could prove too big a step in their development time frame.
For example, will alternative interfaces be good enough in just four years to permit cars without standard controls? After all, many of us have experienced the annoyance of a voice-activated assistant like Siri completely misunderstanding a request to check the weather or a restaurant menu.
If your car simply refuses to understand your destination request or need for a brief stop, it would be significantly more irritating. We’ll expect an autonomous car to perform as if it has an alert and responsive chauffeur, but may end up getting a surly cab driver with limited language skills.
What’s more, even if autonomous cars handle streets and highways with aplomb, how well will they deal with emergency detours or traffic cones? Will they be able to navigate a fast food drive-through or impromptu dirt car park?
To be clear, these situations and more will at some point be easily managed by their software, but will that kind of contextual awareness be standard by 2021? And will we trust it?
The desire of some manufacturers to leapfrog level 3 may reflect the strength of their software. Either way, Ford and Waymo – as we must now call Google’s self-driving car project – may soon be asking the public to put a great deal of faith in new technologies.
Reacting to Ford’s sleeping test drivers, Waymo CEO John Krafcik goes so far as to wonder whether level 3 autonomy “may turn out to be a myth”. We are about to find out.
