The Responsibility of Autonomous Driving

EU’s SARTRE project frees motorists from operating cars.

A new European Union project under the name Safe Road TRains for the Environment (SARTRE) was introduced in October that not only offers improved fuel efficiency and greatly increased vehicle autonomy, but also frees the motorist from operating the vehicle over long stretches of highway driving.

Autonomous control of vehicles is now conducted only by a few Advanced Driver Assist System (ADAS) systems and almost always is implemented as an emergency preventative or corrective measure. Collision mitigation systems can employ full braking power to lessen the severity of an imminent accident, while lane departure prevention can correct steering to avoid leaving a lane. Each of these situations becomes more dangerous at highway speeds, making the speed of preventative action more important. This is why ADAS solutions tend to have the edge over humans.

In a few systems available today, the driver can voluntarily relinquish partial control of a vehicle to lateral movement while letting the ACC system control following distance and speed. Autonomous parking solutions also allow the driver to submit control to vehicle systems, albeit at much lower speeds and therefore with lower risk. But at what point does autonomous control become less safety-centric and more of a bonus to the driver’s comfort and convenience, simultaneously removing the burden of driving from the driver and placing it on the metaphorical shoulders of a computer? The examples listed above can help illustrate this difference.

Collision mitigation systems are safety-centric and emergency-use scenarios, whereas ACC and autonomous parking are selective and are aimed at reducing the driver’s workload, with minimal benefit in terms of safety. Yes, ACC can now slow the vehicle to a stop in addition to smoothing out highway driving, but the former can be accomplished by a collision warning system while the latter is arguably aimed at green driving at bes—or driver convenience at worst.

The Evolution of ADAS
It is precisely this distinction that deserves attention from the industry. ADAS solutions initially were conceived and found their place as safety applications designed to mitigate the effects of an accident or to help drivers avoid accidents in the first place. Anti-lock brakes are an early example of these types of systems.

More recently, ADAS has begun to provide relevant information to drivers to aid them in making correct and informed decisions. However, as technology has improved, the scope of ADAS has begun to expand yet again. ADAS today still includes safety and informational systems, but it now also can be used to address more domains, such as eco-friendly applications and driver convenience.

ACC, for example, can have fuel-saving benefits for all vehicles on the highway, if as few as one in 10 vehicles use the system. However it seems that ACC, as an example, is often embraced as a convenience option as much as it is safety- or eco-related. As more of today’s available ADAS moves into the realm of selective autonomy, and therefore ADAS research moves closer to full autonomy, the industry risks losing sight of safety in favor of relieving the driver of the burden of operating a 4,000 pound or heavier motor vehicle.

However, as technology evolves and we become more fascinated with “cool tech,” it’s important to keep a clear focus on the purpose of ADAS: assisting drivers, not replacing them. This distinction isn’t always easy to define. The EU SARTRE project is an interesting example of this quandary.

SARTRE
The SARTRE project involves a lead vehicle that is fully operated by a sentient and knowledgeable driver who is equipped with a communications system that allows other similarly-equipped vehicles on the road to connect, follow, and form a road train along the highway. The lead vehicle drives the route, communicating speed and other parameters to the following vehicles, which automatically adopt these parameters and autonomously trail the lead vehicle, leaving the drivers without the burden of driving until they choose to exit the road train.

SARTRE expects to provide many advantages, including eco-related, safety and convenience benefits. By autonomously following the lead vehicle in a line, with as little as a few inches between vehicles, every follower exploits the resultant lower air drag and removes the oft-inefficient acceleration by humans, thereby increasing fuel efficiency and lowering vehicle emissions and pollution.

SARTRE also confers safety advantages by taking the human factor out of decision-making while following. Autonomous following also allows drivers to spend their morning commutes performing other activities, such as reading the newspaper, enjoying breakfast or getting a head start on the day’s work. Drivers presumably could retake control anytime they wish, but with the vehicle on auto-follow, the SARTRE project expects to substitute drivers for long stretches of highway driving.

Certainly, the SARTRE project is an example of the advanced state of technology and our understanding of it. SARTRE projects that it will provide multiple benefits, in areas including safety, environmental and convenience. But if the consumer sees the solution as a relief of responsibility, and that’s precisely what happens when the vehicle assumes control, then this presents not only a problem but an opportunity for discussion as well.

Conclusion
ADAS continues to evolve and tackle more complex situations. The argument can be made that a computer reacts faster than a human in similar conditions, and that’s precisely what ADAS has achieved in areas such as collision mitigation and emergency braking.

However, in a time when drivers are more distracted and mobile devices provide more information and entertainment than ever before, the industry must be aware of the consumer’s thoughts and assumptions. If SARTRE removes the driver from the act of driving, will the driver pay any attention to the road at all? If a train of eight SARTRE vehicles, spaced inches apart, is suddenly cut off, is there enough time for a computer to prevent eight accidents?

Unfortunately, there are no answers for all these questions—yet. This is precisely why these levels of autonomy are still in the research phase. It’s logical to assume that technology will continue to evolve and that one day our vehicles will drive themselves to work and back.

However, until the ADAS of the future appears in showrooms, these questions must be asked early on in the development phase with the expectations and behaviors of drivers in mind. In the meantime though, we must be confident that the same bright minds that conceptualized these solutions are also tackling these most difficult questions as well.