SORRY FOR THIS recurring fixation on autonomous cars. I’ve already questioned several practical aspects in ”Autonomous Vehicle Worries” and ethical ones in ”Autonomous Kills.
These are put in perspective by the National Highway Traffic Safety Administration’s categorization of what it terms Automated Vehicles. NHTSA defines five levels of vehicle autonomy:
Level 0) No automation.The driver is in complete and sole control of the vehicle. Many cars on the road today operate in Level 0. In the infancy of the automobile, this would have included such things as throttle mixture and ignition spark control. Until the advent and eventual ubiquity of ABS, braking had no electronic aspects.
With a 1920 Rolls-Royce Silver Ghost, Level 0 included lots of driver functions. Image from eveningnews24.co.uk.
Level 1) Function-specific automation. Today, Level 1 includes things like ESC (Electronic Stability Control) and pre-charged braking (for vehicle assistance in stopping in a reduced distance). These are increasingly common features. In fact, NHTSA has proposed that ESC be standard on some new vehicles.
The sensing elements of Electronic Stability Control. Other actuating elements evoke corrective control. Image from the Insurance Institute for Highway Safety.
Level 2) Combined-function automation. Level 2 establishes interaction of automated features. Cruise control with collision-avoidance braking is an example. Such interactions are not uncommon today as premium features on some cars.
Level 3) Limited Self-Driving Automation. Level 3 assumes full control of the vehicle, though it requires the driver to take over, in NHTSA’s words, “with sufficiently comfortable transition time.” That is, it would be fitted with full driver controls.
This sci-fi car would likely qualify as Level 4. Clearly its driver isn’t prepared to take control. Image from rand.org.
Level 4) Full Self-Driving Automation. With Level 4, the vehicle performs all functions and monitors conditions for the entire trip. A “driver” merely provides destination and navigational input. Indeed, there may not even be a steering wheel or other controls.
Among automakers and other committed parties (Google, Apple and Uber, to name three of the latter), there are differences of philosophy. Ford, for example, is pursuing a direct move to Level 4, with cars for ride-hiring and ride-sharing on the road in five years. By contrast, GM, Nissan and Volvo plan to introduce Level 3 vehicles as a transition to the ultimate autonomy of Level 4.
Volvo XC-90 autonomous vehicle entering a Swedish test fleet.
A major reason for Ford’s decision lurks in the word “transition,” as it appears in the NHTSA document. What’s “sufficiently comfortable”? How confident can an automaker be, and what’s the corporate liability, that the driver will successfully react when the vehicle gives up control?
Even with a fully functioning Level-4 car, there are potential challenges a’plenty. Automotive News, September 5, 2016, cited two. Staff Reporter Richard Truett’s piece is titled, “To Get Driverless Cars, Fix the Roads.” He cites a Department of Transportation study that details our crumbling infrastructure. Human drivers can avoid potholes and overlook missing lane markings. But what about their computer-guided alternatives?
What’s the optimal protocol of an autonomous car/pedestrian interaction? Image from Automotive News, September 5, 2016.
Another article, “Street Smarts,” by Bradford Wernle, discusses the Level-4 car/pedestrian encounter. Human drivers (or at least the considerate ones) try to establish a pedestrian interaction at intersections or crosswalks. Maybe it’s a nod of the head, or a friendly wave.
What’s a Level-4 car to do? Among the options are a flashed emoji, a blink of the headlights, anything but a monotone horn blare.
“Hey,” says the Level-4 car, “You’ve got pretty eyes.” ds
© Dennis Simanaitis, SimanaitisSays.com, 2016