- Aquaculture, Equine & Livestock
- Architects & Engineers
- Aviation & Aerospace
- Consumer Goods & Services
- Education & Public Entities
- Entertainment & Leisure
- Financial Services
Safer Roads in a Driverless World?
April 11, 2014
Our Future: Giles, the Driverless Car
Giles pulls up to the curb. He scans your body to verify your identity and gauge your mood—the right music makes the commute more relaxing. His door opens, and out pipes your favorite Beatles song. You bark out your destination, jump in, throw your bag on the floor, sit back, and start tweeting about your date last night. Giles buckles you in, closes the door, and weaves smoothly through rush hour traffic. He gently pulls up to the curb in front of your office, opening the door. You hop out, the door closes, and Giles zips off to his next, shared-mobility passenger.
Rewind! How did we get here?
The Need for Speed
“Sheer driving pleasure!” is the BMW slogan. Until now, driving has not only been exhilarating, it has also helped us conquer distance. Mustangs, chariots, boats, cars, airplanes and rockets! Since the earliest horsemen 6000 years ago, humans have used their wits to harness the acceleration of other creatures, and then of technology, to reach the far ends of the earth and into space.
The command of this speed has spawned thriving industries, the greatest arguably the automobile. Since its invention at the turn of the 19th century, the car has quickly evolved into an international symbol of status, individuality, freedom and power.
Disruptive Technology Disruptive technology is innovation which upends an industry, engenders an entirely new market, or fundamentally alters the way society operates.
Computers, smartphones, and email have transformed the cultural and corporate landscape, and now startling new technologies are springing up in every arena.
No one could forecast the full outcome of these innovations, but we have some ideas. Welcome to the third part in our Fast Fast Forward Disruptive Technology series.
But the image of a man behind the wheel of a convertible, arm around the wife, kids in the back, wind in their hair, is a retro dream. Our global populace has reached the tipping point—we don’t have enough open space for everyone to enjoy his own private, speed heaven anymore.
Shrinking Room for Human Error
There are just too many of us. The global population has grown by a billion inhabitants every 12-14 years since 1974, and the United Nations predicts an increase from 7.4 billion now, to 8 billion by 2027.
On a packed urban highway, a careless driver can turn a car into a weapon. According to the WHO (World Health Organization) 2013 Global Status Report on Road Safety, 1.24 million people die on our roads annually. Another 20 to 50 million people are injured.
Fully 95% of the time, human error is to blame, and it isn’t just car drivers and passengers who die. Half of the victims are what the WHO calls “vulnerable road users”: 23% motorcyclists, 22% are pedestrians, and 5% bicyclists.
Everything Else Has Failed
High on the thrill of speed, humans have a knack for forgetting about the car’s superhuman physics.
Cars feel like natural extensions of our own bodies. Unfortunately, they’re not, and our reflexes aren’t what we think they are. Even athletes in top condition enjoy less than a split second’s reflex advantage over short-sighted, less fit, middle-aged drivers. Scientific studies have proven that drivers wildly overestimate their reflexes, and have no real concept of vehicle momentum (force multiplied by speed). In other words, we cannot intuitively gauge how long it takes to slow or stop a car.
The rising global middle class, the spread of private car ownership, and the ensuing mayhem on the roads have caused us to tackle safety and congestion issues from every conceivable angle over the past few decades: speed limits, seat belts, anti-lock brakes, air bags, new outer airbags to protect pedestrians, crumple zones, and roll-proof frames. All of these have dramatically reduced death and injury, and there are many countries where tragedy could still be mitigated through such measures. But it is not enough.
How Much Safer Could a Fully Autonomous Car Be?
The car’s attention never wanders. The car receives real-time data on events occurring around the next bend, invisible to the human eye, and begins responding and compensating immediately. History shows that humans learn from mistakes. The collective conscience of autonomous cars, constantly accruing knowledge and learning to adapt to every nuanced traffic condition or calamity, will inevitably exceed by magnitudes the limited experience of a single human driver. Since 2010, Google has accumulated 700,000 kilometers of evidence showing that its driverless software alone, integrated into a standard Prius or Lexus, beats professional human drivers. The drives are smoother and safer.
Now consider equally sophisticated software, combined with fine-tuned physical adaptations and multiple navigational sensors, including cameras, laser imaging, radar, and ultrasound. Sensors and computers will work with GPS to detect the environment, eventually down to the centimeter.
Incrementally Edging Out Human Error
Since 2010, Google and the media have done a great job of massaging the idea of a totally driverless car into the public conscience. But carmakers are more inclined to offer new, super-automated features as part of a historical trajectory of improved safety and comfort. Words like “driverless” or “autonomous” are still offensive to some car owners, for whom driving is the number one pleasure. Their keys will have to be pried from their fingers, before they’ll let a car do the driving for them.
The increasing range of autonomous features being sold by Mercedes, BMW, VW, Audi, and Volvo are wisely dubbed “advanced driver assistance” programs: GPS with integrated, traffic updates, adaptive cruise control, brake assistance, lane-departure warnings, lane-change assistance, collision-avoidance, night vision, attention assistance, adaptive light control, blind-spot detection, hill-descent control, traffic-sign recognition, automatic emergency calls, parking assistance, and more.
Oh yes, these cars do the parallel parking for you: Inspector Clouseau and the owners of his many dented victims breathe a sigh of relief.
But That’s Not All!
At the CES (Consumer Electronics Show) in Las Vegas this January, BMW demonstrated exactly what its 360-degree radar, cameras, and ultrasonic sensors, plus both lateral and longitudinal control units, could do in dire conditions. Its driverless, 6-Series Gran Coupe blew the minds of its passengers and observers. It took slaloms at racing tempo, and kept the pedal down on a watery bend, where it automatically counter-steered through a sideways slide, to keep from doing a 180.
BMW told the CES that they would like to do further testing in the most extreme conditions. Not to verify that the cars can perform up to human driving standards. The tests are to make sure they are considerably better.
There is bound to be a long, transitional period, a decade or two of autonomous cars mixing with human-driven cars. But human drivers may increasingly feel like outsiders on a road mastered and dominated by exquisitely navigated, driverless cars.
Remove the Human Error: 95% Accident Reduction
Once all of these “advanced” features are integrated into a holistic, driverless system, it won’t be long before autonomous cars are the only cars on the road, at least in cities. If the car systems don’t fail, without human error, we’ll be looking at a nearly 95% reduction in accidents.
But what happens when an electronic surge or hacker sabotage inflicts a massive system outage? All the car systems in an area could be knocked out, and the drivers forced to take control. Currently automated systems require a human to take control in certain scenarios. Eventually, that might not be the case. In its September 2013 paper on autonomous vehicles, the UK Parliamentary Office of Science and Technology cited the inevitability of hacker attacks, and proposed a driverless system designed to “fail safely”. How exactly such a system would work will have to be spelled out later. It would be extremely frustrating not to be able to take control as a driver. On the other hand, it could be chaotic if we did.
The Meaningless License
Why is that? Because licensed drivers might be so inexperienced by then, that they won’t be able to cope when the system fails.
It seems inevitable that we will witness the same skill deterioration in drivers that we have seen in commercially certified pilots: all theoretical knowledge, reduced real-life skill. Thanks to the magic of autopilot, today’s pilots only pilot airplanes for about 2.5 minutes of long-haul flights. When the sophisticated auto-pilot system fails, look out, because younger pilots may not be able to cope as well as their predecessors.
Having a driverless “fail safely” protocol that works will be critical.
Clean Air in the City?
Driverless cars could represent a turning point in greenhouse gas emissions.
71% of emissions in the EU come from road transport. So, in addition to battery-operated cars, the EU considers “smart cities” and driverless cars a very real part of the emission-reduction solution.
First, every car-train, or platoon, experiment has proven that fuel efficiency is improved up to 30% when cars travel in regimented, streamlined groups. Braking and accelerating are reduced, along with greenhouse gas emissions.
Second, a shift from private ownership to shared mobility would also decrease the number of cars on the road by up to 900%, a huge emission reduction.
How Close Are We?
In an interview with the BBC Top Gear team, Mercedes acknowledged that the technology required for a fully driverless system is, in fact, already included in the 2014 S-Class.
In That Case, what Are We Waiting For?
Infrastructure? Not really. Many people don’t realize how close the infrastructure is to handling driverless cars, en masse. The data is ready. We have GPS, mobile networks, and highly detailed, digital maps. And soon all of our roads, junctions, signs and signals will be digitally plotted.
But right now, the vehicles are doing all the communicating and data collecting. Those few driverless cars on the road are managing very well—better, in fact, than the chaotic human drivers around them. But ultimately, stoplights, signs, lane markings, and other traffic features will communicate with vehicles, and with government agencies.
What About Those Vulnerable Road Users: Pedestrians and Bicyclists?
This is a big question. Even if we grant that the cars and infrastructure will be in continuous, fail-safe communication—a big concession—who will be communicating with the pedestrians and cyclists? Will Giles, the driverless car, possess a humane hierarchy of priorities? If the choice is between hitting a pedestrian, sideswiping a cyclist, or hitting another car, what will Giles do? What about the cat crossing the road?
The Legislation is Not Quite Ready
Despite the open questions, the EU and many national governments are preparing for the inevitability of a driverless future, and contemplating regulation. In The UK believes a driverless cityscape is certain, and that it will herald unparalleled traffic coordination, managed through vehicle-infrastructure communication. Some speculate that speed limits will even become obsolete, when cars can intelligently select the safest speed for certain road segments.
Currently only a few jurisdictions explicitly allow driverless cars. In the USA, only Nevada, California, and Florida allow them. The EU continues to introduce incremental legislation paving the way for them. As of 2014, cars without automatic emergency brakes will not be awarded a five-star safety rating in the EU. And in 2015, basic, embedded SIM cards, which automatically dial the emergency center in a crash, will become standard in the EU.
Currently the Vienna Convention requires all vehicles to be manned by a licensed driver. This may not change as soon as driverless cars are available, but it could in a decade or two. Then certain groups would enjoy newfound independence: seniors, children, the blind, and the infirm. Could we create special vehicles or reserve special lanes for them? What about alcoholics and drug users? Should they be rewarded with similar liberty? Should we ever permit a driverless car to take to the road, without a sober, competent human behind the wheel, in case of emergency?
The Key Question: Liability
The main question legislators face now is: Who should be blamed when driverless cars crash, especially in the first two decades, with human drivers still have their hands on the wheel?
Several key automakers are working with California legislators to draft detailed driverless legislation, to be presented for debate in January 2015. The bill is expected to take a balanced approach to liability, and take into account the safety benefits of driverless cars. If they follow through, this will be the first serious legislation on autonomous cars, and is likely to set the tone for laws worldwide.
Lines will have to be drawn in very specific contexts. Will the “driver” ever be to blame? Would it be possible to misuse or abuse a driverless car and render it unsafe? For example, what if a negligent driver fails to take the car in for repair? Warranties will surely need to include repair and usage clauses.
What about mechanic liability? Mechanics will need sophisticated electrical and digital expertise, and continual skill training. The wrong replacement part or faulty workmanship could easily cause an accident.
Automakers and suppliers could bear much greater liability risk. Manufacturing, operational, and supplier standards will have to be incredibly high, to avoid catastrophe, devastating financial loss, and death.
How Will Insurers Cope?
The “black box” is going to solve a lot of problems. Until now, in a mass collision, insurers have had no real way of discovering the root cause, and have been compelled to divide up the total loss and pay out. This is going to change.
Google’s driverless cars, and other cars with advanced “assistance” programs, already have black boxes, like airplane black boxes, which record digital and mechanical data, as well as any biometrics available at this stage. As the data becomes more and more detailed, it will get easier and easier for police and insurers to reconstruct the events leading to an accident, and pinpoint the cause, from faulty software and components to driver error.
Insurers are already charging lower premiums to owners of highly automated cars, and many brokers offer discounts to drivers with black boxes. Manual driving is going to get more expensive in contrast.
Risk modeling will have to change dramatically. Today’s frequent, low-damage collisions will be history, but a new category of ultra high-risk accidents will emerge. The worst-case scenario: hackers inflict a city-wide, system blackout; and panicked, unskilled drivers either can’t cope, or can’t even take control of the cars. The result: mass collision. Automobile cyber terrorism is likely to be a requisite part of future insurance policies.
Another niggling problem remains: The idea that a driver would not even have the option of taking control in the event the system failed. If the steering, the gas, or any other primary function went berserk, would the driver be forced to watch helplessly as he and his family drove off a cliff, or into a head-on collision? Even a one-in-a-million error would be unacceptable. This is a problem carmakers must solve.
The Orwellian Nightmare
In addition to losing our driving privileges, we face a new loss of privacy. Our smart phones already supply police, corporations, hackers, and spies with our location, business and personal communications, and web-browsing and buying habits. All this, plus, eventually, our heart rates, blood alcohol levels, video footage, and other information, will become part of a continual surveillance feed. We have to accept the fact that cars will become public spaces. And as the CCTV (close-captioned television) signs throughout the UK remind us, we should expect to be observed, by authorities.
New Manufacturing Business Model
The carmakers have created this technology. So what will the new business model be? Some suppose that precise traffic coordination will make room for more private vehicles. That might be true, but when people no longer get to revel in the “sheer pleasure” of actually driving, private ownership is going to be a lot less attractive.
That’s why it’s more likely that drivers, or riders, especially in cities, will turn to vehicle sharing. French and American studies have already observed this trend growing over the past decade, especially among young, urban workers.
The Future: Greener, Safer, and More Comfortable
In a decade or two, we’ll probably hit the “antique” car lot, to pay for the “sheer pleasure” of driving for an hour or two in exclusive, cordoned-off territories. We’ll have to sign liability waivers, of course.
One thing is certain: we won’t be going backwards. The cars are almost ready. We have almost all the geographical and traffic data we need. The communications technology is ready, just waiting to be installed in cities. And legislators are making plans. In as little as ten years, driverless cars could be the smooth, silent, and safe majority on our roads.
The future may be slightly less thrilling, but it will also be a lot more relaxing and comfortable. And there will be new thrills. While we recline, read, text, or watch TV in our shared cars, we’ll be saving the planet, reducing emissions by at least 30%. We’ll also be saving over a million human lives every, single year, and preventing up to 47 million injuries. Maybe we’re really just trading one pleasure for another.
We at XL are preparing for that day.
Harald Knothe is Insurance Client and Distribution Leader for XL Group in Germany.