This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.

Post

On the Cusp of a Driverless Transition?

Sussex Energy Group's picture
Sussex Energy Group

Creating sustainable energy systems will be a defining challenge for humanity in the 21st century and one that requires an understanding of the technological, economic and political dimensions of...

  • Member since 2018
  • 52 items added with 35,747 views
  • Feb 2, 2018
  • 872 views
autonomous

Nissan Autonomous Drive Vehicle. Image by Norbert Aepli. Shared under Creative Commons Attribution 3.0 Unported license.

There is a great deal of excitement about the potential of driverless vehicles (also called self-driving, autonomous, connected and autonomous (CAV)) to contribute to a safer, cleaner, and more efficient and equitable transport system. New entrant (e.g. Tesla) and incumbent (e.g. Volvo) vehicle manufacturers, mainstream news media, the government and even the Queen are all leading discussions of the UK’s role in the design, development, demonstration, manufacturing, and use of driverless technologies.

With the international driverless vehicle market currently growing at 16% a year and a suggested value of £900bn by 2025, the UK government is eager to be a key player. Moreover, in a post-Brexit context, automation and robotics are central to the stimulation of domestic industry, articulated in the 2017 Industrial Strategy.

The government has undertaken a range of steps to help stimulate domestic innovation, and nurture an environment conducive for on-road trials. This has included reviewing road rules and regulationsfunding research, and establishing the Centre on Connected and Autonomous Vehicles. This culminated in the new Automated and Electric Vehicle Bill announced by the Queen at the 2017 State Opening of Parliament. The Bill is designed to “allow innovation to flourish and ensure the next wave of self-driving technology is invented, designed and operated safely in the UK”.

Public demonstrations and trials have been central to the emergence of driverless vehicles in the UK, and elsewhere (e.g. Germany, Norway, USA). Such demonstrations are increasingly focused on the implications of automated technologies on infrastructure (e.g. transport, ICT), insurance, user behaviour, policy and regulation. Since the first wave of demonstration projects in 2015, trials have grown in spatial scale(s), the variety of technologies tested and empirical focus.

Despite current hype around the novelty of driverless innovations and their ability to radically alter the current system of mobility, driverless technology has a relatively long history. Since the rise of private vehicle ownership in the 1920s there has been interest in innovations that reduce or remove the role of humans from the driving task. Predictions of contemporary transport concerns (including congestion) motivated a rethink of the role of human drivers in the system of motorised transport. Yet for a variety of social, economic and technological reasons, support for trials and demonstrations eventually weakened, until recent developments in deep machine learning and other (groups of) ‘smart’ technologies enabled new forms of driverless technologies to emerge.

A range of categorisations have been developed to enable technological and policy understanding of automation. The National Highway Traffic Safety Administration (NHTSA), the Society of Automotive Engineers (SAE) and Bundesanstalt für Straßenwesen (BASt) have each developed ways of framing degrees of automation which distinguish between vehicles which largely rely on human control (SAE levels 0-1), to those which are to a large degree (SAE level 4) or completely (SAE level 5) operated by a machine (Fig. 2). There is a commonality across the definitions, each articulating a distinct set of roles for the human driver, and for the machine(s).

automation

Source: Author’s own depiction of SAE levels of automation from a human driver perspective.

The SAE levels, while not without problems, are useful for signalling the various stages of automation, and help us to understand the automated features in contemporary vehicles. Features including lane assist, park assist, and area view are becoming increasingly common in new vehicles.

The key word here is ‘assist’, where automated features are currently assisting the human driver, who has responsibility for the driving task. As the automated capabilities increase, the responsibility incrementally shifts from the human to the machine. Yet the spate Tesla Autopilot collisions have pointed to public and industry confusion in interpreting the degrees of automation and fully articulating their capabilities ‘on the roads’ and in real-world settings.

Nevertheless, in the public sphere, awareness and knowledge about automated technologies and degrees of assistance is growing. This is playing out in new car advertisements with the promotion and normalisation of driverless ‘assistance’ innovations. For instance, the 2016 Volkswagen Tiguan advert titled “Parents, Take Note depicts parents and teenagers at the school gate as the driver of the Volkswagen Tiguan lifts his hands from the steering wheel as the vehicle parks itself in front of the school, to the awe of those watching.

Powerful and vocal actors are enthusiastically supporting the development and testing of various levels of automated technologies and full driverless vehicles (SAE level 5) in the UK. Yet replacing current vehicles for automated vehicles is unlikely to result in the claimed energy, safety, efficiency, and equity benefits. It is critical that discussions of automation broaden to include consideration of ownership models, and the potential for shared mobility, alternative fuels, infrastructures, urban form, and user practices.

By Dr. Debbie Hopkins, Centre on Innovation and Energy DemandTransport Studies Unit, University of Oxford

This blog draws on the work of the ‘The energy implications of automated and smart freight mobility‘ project.

For further reading please see: Hopkins, D. & Schwanen, T. (2018). Governing the race to automation. In: Marsden, G. and Reardon, L. (Eds.) Governance of Smart Mobility. Emerald, Bingley, UK.

Original Post

Discussions
Bob Meinetz's picture
Bob Meinetz on Feb 4, 2018

Debbie, I took the liberty of editing your illustration to reveal a glaring blind spot (pun intended) in the assumption automating an activity with lethal, public dangers might serve society. With the conceit of “driverless” cars, we promote convenience at the direct expense of responsibility:

http://www.thorium-now.org/images/zeroresponsibility.jpg

Imagine a pedestrian hobbled by a driverless car, which didn’t “see” him in a crosswalk. (no one doubts this situation will occur, and that driverless cars might be safer statistically is irrelevant to him and his family). He is asking, “Who will pay for my physical therapy? Who will compensate me for my lost wages? Who…” The answer is in the lower right corner of the illustration: no one.

Those who suggest a carmaker might be liable, a software or mechanical engineer might be liable, ad infinitum, are people who have never filed a lawsuit against a multi-$billion company.

Are we on the cusp of a driverless transition? Of course not. Commercial aviation has had the capability of 100% automation for years, yet Boeing and Airbus know no one would fly in planes piloted by a bot – and there’s a lesson in that. It would be an unspeakable tragedy if it took hundreds or thousands of highway casualties to learn the same lesson for cars.

Jesper Antonsson's picture
Jesper Antonsson on Feb 5, 2018

With no liability, that would be another factor that could help drive adoption of autonomous cars. And if the car owner is required to have insurance that covers accidents, then the statistics is quite relevant to determine the insurance fee.

Engineer- Poet's picture
Engineer- Poet on Feb 5, 2018

Imagine a pedestrian hobbled by a driverless car, which didn’t “see” him in a crosswalk. (no one doubts this situation will occur, and that driverless cars might be safer statistically is irrelevant to him and his family). He is asking, “Who will pay for my physical therapy? Who will compensate me for my lost wages? Who…” The answer is in the lower right corner of the illustration: no one.

Hogwash.

States will insist that driverless cars carry insurance just like human-directed cars.  The insurance will pay the pedestrian’s losses and medical bills.  If the driverless car is safer, the insurance premiums will be lower.  If a fleet operator is big enough to self-insure to save the overhead costs (e.g. GM now owns Lyft), then that’s what will happen.  They’ll have to conform to state insurance laws, just like the huge, very-hard-to-fight insurance companies do.

It would be an unspeakable tragedy if unreasoning fears kept us from saving thousands of lives and preventing untold numbers of injuries.

Bob Meinetz's picture
Bob Meinetz on Feb 5, 2018

EP, this is funny – who will pay for the insurance, Mr. Driverless Car? Will it go onto Mr. Car’s record so his rates will go up? Will there be any consequences if Mr. Car’s actions can be considered reckless homicide – will you throw Mr. Car into jail? Who will represent him, the legal firm of Bot, Bot, & Bot?

I hope you nor anyone you care about is injured in the Brave New World Where No Human is Responsible For Anything, because “thousands of lives” are nothing more than a convenient abstraction until it’s someone you know. That’s not fear, that’s reality.

Bob Meinetz's picture
Bob Meinetz on Feb 5, 2018

Jesper, statistics are insignificant compared to a driver’s record in determining insurance fees. Shall we have a corresponding statistical payout – every victim gets the same compensation, whether their kid got run over or their bumper got dinged? It’s not the car owner’s fault your kid was a statistical anomaly.

Mark Heslep's picture
Mark Heslep on Feb 5, 2018

The issue of driver-less vehicles is more complicated than supposed.

By far the largest single source of fatal vehicle accidents in the US still stems from impaired drivers (alcohol), about a third. With automation, those fatalities will fall drastically. On the other hand, automation flaws leading to fatalities are inevitable (though relatively fewer), and will occur across random driving situations, including those in which humans currently have their highest attention levels, such as taking the kids to school or, as others have said, pedestrians at typically safe cross walks, where the public tolerance for occasional fatalities is near zero.

Jesper Antonsson's picture
Jesper Antonsson on Feb 6, 2018

Jesper, statistics are insignificant compared to a driver’s record in determining insurance fees.

With autonomous vehicles, the driver’s record will be the statistics (for a particular vendor).

Shall we have a corresponding statistical payout – every victim gets the same compensation, whether their kid got run over or their bumper got dinged?

Don’t understand what you’re getting at here. The payout is meant to cover/compensate for damages. Is the damage different, the payout will be different.

It’s not the car owner’s fault your kid was a statistical anomaly.

No, but he’ll have to carry insurance to cover the risks inherent in operating a car. I don’t see the problem.

Jesper Antonsson's picture
Jesper Antonsson on Feb 6, 2018

who will pay for the insurance, Mr. Driverless Car?

The owner.

Will it go onto Mr. Car’s record so his rates will go up?

Yes.

Will there be any consequences if Mr. Car’s actions can be considered reckless homicide – will you throw Mr. Car into jail?

The vendor’s leadership and engineers might, if there’s sufficient evidence. That happened to Volkswagen’s staff that were involved in the diesel emissions cheat.

I hope you nor anyone you care about is injured in the Brave New World Where No Human is Responsible For Anything, because “thousands of lives” are nothing more than a convenient abstraction until it’s someone you know.

If I have to choose between the satisfaction of revenge if my kids are injured and lowering the risk that they are injured to one tenth, I’ll choose the latter.

Engineer- Poet's picture
Engineer- Poet on Feb 6, 2018

who will pay for the insurance, Mr. Driverless Car?

You realize that cars are chattels, not people?  The car’s owner pays.

Will it go onto Mr. Car’s record so his rates will go up?

Yes, of course.  Any software suite with a higher rate of accidents will result in higher insurance rates for vehicles using it.

Will there be any consequences if Mr. Car’s actions can be considered reckless homicide – will you throw Mr. Car into jail?

Machines cannot be reckless, but humans who roll out badly-designed or ill-tested machine software can be.  You can fine or jail the management of the company as well as the company itself.

Of course, humans can be malicous, and computers can be hacked.  It wouldn’t be too long before the first murder case where the weapon was a hacked autonomous vehicle.  The mfgr would be liable for negligence if their hacking defenses weren’t up to snuff.

Who will represent him, the legal firm of Bot, Bot, & Bot?

Given how software is taking over legal work, this is not as far-fetched as you think.  I wouldn’t be surprised if fault would be assessed from the recorded video of the car’s cameras and payment made with the only human intervention being a review of the evidence.  This could be done in days or hours instead of years and cost very little per case.

Will you stipulate to the fact that a driverless car with half the accident rate of human drivers would be (a) safer for humans and (b) cheaper to insure?

Bob Meinetz's picture
Bob Meinetz on Feb 6, 2018

EP, I will stipulate that a driverless car with half the acident rate of human drivers falls under the heading of unverifiable hype, akin to what I’ve been hearing for at least the last four decades.

It’s a shame you feel compelled to introduce a faulty human to review a software ruling…couldn’t a software review do the job much more efficiently? Or maybe, we make humans responsible for determination of guilt or innocence, EP, because responsibility is, and will always be, beyond the scope of digital intervention.

Software is taking over legal work, is it? You think it’s only a matter of time before an automated attorney is giving closing arguments to a human jury? Might as well automate the jury and the judge too…what could possibly go wrong?

This is fun.

Bob Meinetz's picture
Bob Meinetz on Feb 6, 2018

Jesper, you confuse revenge with accountability.

One tenth? Check with EP, he seems to think risk to your kids is only one half better than human drivers.

Tell me rolling the dice was the right decision when your kid is lying in the street, and the bot-driven car which hit him/her is on its way to pick up its next fare.

Jesper Antonsson's picture
Jesper Antonsson on Feb 8, 2018

You specifically asked if we could throw Mister Car into jail, and what is accountability, really, beyond a mix of revenge and compensation? Compensation will be fixed by insurance, so your remaining complaint is that you might not be able to exact revenge. Autonomous cars are safer (doesn’t matter exactly how much), so basically, your thirst for revenge trumps any instinct you might have to reduce harm. You should think hard about whether that is a civilized approach.

Btw, when I say nuclear is safer than other energy, a lot of people unreasonably respond “tell that to the citizens of Chernobyl/Fukushima”. But yeah, nuclear is still safer. That you can find a few victims of it isn’t very meaningful, regardless of accountability.

Sussex Energy Group's picture
Thank Sussex for the Post!
Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.
More posts from this member

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »