Automotive > Infotainment & Telematics Blog



Tesla Has Auto Industry Facing the Autopilot Blues

by Roger Lanctot | Dec 07, 2015

Tesla’s launch of autopilot on the Model S P85/P90 has the auto industry singing the blues. Few auto makers are prepared to launch similar functionality in their own vehicles. But if they do or if they try, they are walking a technology plank intended to eliminate drivers and possibly vehicle ownership.

These thoughts have descended on me ever since my first ride in the Tesla Model S with autopilot. I realized over time the auto industry can’t do it.

This perspective has been reinforced by observers engaged in the process of automating flight. As more flight tasks are automated, pilots are less attuned to the skills required for previously manual tasks.

As automation reaches its limits, pilots increasingly find themselves asked to retake control of tasks at which they are no longer adept. The good news is that in flight there are multiple redundant systems and normally ample time to recover.

(I refer you to Dr. Steve Casner's excellent work for NASA: http://tinyurl.com/zrbmx8f)

This is not true in a car. For this reason, experts in human factors in the airline industry suggest that Google is on the correct path (no steering wheels and no pedals) in working toward complete autonomy – what the industry and the National Highway Traffic Safety Administration refer to as Level 4.

The auto industry is currently flirting with Level 3 automation which combines adaptive cruise control with traffic jam assist and lane keeping to allow the car to control steering, acceleration and braking – but with the expectation that the drive remains attentive. By launching autopilot as a beta with risk assumed by the driver, Tesla has poked its finger in the eye of Level 3’s NHTSA definition.

For their part, airline human factors experts have identified too many points of failure in the process of de-activating autonomy and returning control to the pilot – or driver. The process is challenging, but works in airplanes. These experts strongly recommend against applying this approach to cars.

In my experience of autonomy, the owner and driver of the car only took me and a friend on a short ride in the Detroit suburbs. Tesla’s autopilot is only intended for use on highways, but my friend was activating the feature on surface streets. (Youtube videos show that it is capable of functioning admirably in city traffic as well.)

While using autopilot on surface streets the driver of the Model S was able to pass cars by simply activating the turn indicator and letting the car do the rest. There were some unnerving moments when approaching intersections since the car was not capable of recognizing intersections, tuned as it is for highway use.

As I reflected on the brief experience I looked at it through the eyes of auto industry executives, many of whom I interact with on a daily basis and some of whom were outraged at Tesla’s autopilot audacity. It occurred to me in auto-industry-think that the car should identify the driver and determine if the driver was sufficiently familiar with the system, in case he or she were to use the autopilot.

This also brought to mind the thought that a vehicle capable of any automatically piloted driving will need to ensure that the driver is paying attention or at least present. Mercedes solved this problem by requiring a hand on the steering wheel during its autonomous driving mode – a requirement that is a vestige of the Vienna Convention and which was famously subverted by drivers affixing soda cans to their steering wheels as surrogates for a human hand.

In Strategy Analytics’ testing of the 2015 Mercedes S Class, the Distronic Plus driver assistance package enabled a hands-free autonomous driving mode. So Mercedes has decided that you don’t have to keep your hands on the wheel.

(http://tinyurl.com/pqxl66h - User Experience Evaluation: High-Speed Autonomous Driving in 2015 Mercedes S-Class)

But should each driver be identified and credentialed (and trained?) individually? Tesla’s answer is “No.” Once the autopilot has been activated – and the use-this-beta-release-at-your-risk end user license agreement has been accepted - any subsequent driver can use it.

Mercedes’ answer is “If you can figure out how to turn this function on then you can use it.” Strategy Analytics found the Distronic system in Mercedes to be fairly complex in its activation and de-activation aspects.

Tesla has taken the driver-beware approach with simple human machine interface elements. Mercedes has taken an incremental enhanced cruise control approach with complex and confusing HMI elements. Google intends to take the driver completely out of the loop.

The challenge for car makers boils down to human factors. How can car makers enable incremental enhancements and adjustments to human interfaces to enable a safe evolution to what might be called advanced cruise control? And how can car makers manage an evolutionary process which has, as a final objective, the elimination of the driver?

The advice from the airline industry is to not even try. Testing reveals multiple challenges in pilots seamlessly retaking control of airplanes, even when they have ample time to do so.

The auto industry is facing an existential test that has as its goal the redefinition of transportation. If the car is not to have a driver, will it be owned? That is the question.

The good news for car makers is that the process will take decades. And along the evolutionary path to autonomy new, enhanced driving experiences will be enabled.

Perhaps the best news is that car insurers have yet to penalize owners of Tesla’s with autopilot. Insurers have spoken with their terms and conditions – a Tesla with autopilot is no more expensive to ensure – relatively speaking – than a conventional car.

Saving on fuel AND insurance? The Tesla story just keeps getting better and better.

Previous Post: The Death of Radio Traffic Reports Foretold - Prematurely | Next Post: Radio Will Survive Dashboard Clash
Leave a comment