The Economist magazine has hosted a debate about driverless cars between Futurist Paul Saffo and Andrew Bergbaum (director at AlixPartners, a global advisory firm and co-author of the AlixPartners Annual Automotive Study).
The inevitable arrival of autonomous vehicles is tracing a classic innovation S-curve. First is an extended period—typically a decade or two—when technologies gradually ramp up and innovators explore business models. It is a period of small successes and interesting failures. The general public either is utterly unaware of what is afoot, or concludes it is interesting but unlikely to arrive any time soon.
Then suddenly an inflection point is reached and 20-year failures become overnight successes. Adoption soars and once-futuristic curiosities become unremarkable everyday conveniences. All the innovations we take for granted followed this pattern. TV took off in 1951, going from hopeful possibility to 70% of American households in under a decade. In 1989, networked hypertext was a hazy nerd vision that seemed decades away, but the world wide web arrived a year later, triggering the dotcom revolution of the 1990s. Automobile sales went from nothing to 1m cars on American roads in the first decade after 1900.
Will incumbents will hold back innovation ? No! Just ask the record companies which failed to stop iTunes, or Kodak which watched helplessly as the digital imaging technology it invented utterly destroyed the conventional film market. Of course the arrival of autonomous technologies will disrupt the auto industry and eventually may destroy the current ownership model. But the incumbents can no more slow events than King Canute could order the waves to stop rolling in. Their only option is to get on board—or get run down by insurgents.
In my opening statement I asserted that fully autonomous vehicles would arrive by 2030, but I was just playing it safe. The Society of Automotive Engineers puts the date closer to 2025, and Google co-founder Sergey Brin thinks we will get autonomous rides by 2017. I’ll stick with 2030, but honestly, I am certain that I am wrong: autonomous vehicles will arrive much sooner than seems possible today. And when it happens, today’s sceptics and naysayers will rush to tell us how obvious and inevitable it all was.
Andrew Bergbaum Rebuttal
This debate is not about technology, but about all the factors—legal, economic and social—that permit the technology to be adopted by the mass market.
Not all the technology questions have been answered. For example, questions over performance in snow, sleet or hail are still being worked through. And there are many more such issues.
The legal position is still far from clear. It is true that three states in America have passed legislation, however each piece of legislation has many nuances and none yet permit self-driving cars for anything other than testing. In Nevada the tester has to put up a bond of at least $1m to cover any potential liability. Florida and California have both legislated that their respective Departments of Motor Vehicles should come up with rules for self-driving cars—meaning that they do not yet exist.
The truth is that the new legislation in these states still leaves many issues unaddressed. Bodies such as the National Highway Traffic Safety Administration (NHTSA) will need to move considerately and multilaterally to ensure they have covered all aspects of the regulations for a self-driving car infrastructure, particularly when one considers the complications of having a mixed autonomous/driven car network on the roads at the same time. This will not be a fast process.
Market adoption is notoriously slow in the automobile industry, 13 years after entering the market, hybrids still represent only 3% of total new car sales in America and 1.3% in Britain.
I maintain that we are still many years away from seeing fully self-driving cars on our streets in any meaningful numbers.
The statements and claims are sufficiently vague by Paul and Andrew that they overlap a great deal.
Fully automated cars can arrive and not be in meaningful numbers for decades.
Some fully automated cars could be licensed and on the roads by 2017 and tens of thousands or even hundreds of thousands by 2025.
There is no clear definition of meaningful numbers.
It would be good if Andrew indicated when he thought hybrids arrived in meaningful numbers or when he thinks they will.
There are about 3.5 million semi-trailer truck drivers in the US and about 5 million registered semi-trailer trucks. There are about ten million semi trailer truck drivers in the world.
I believe those trucks will become highly automated with platooning so that one skilled driver would be in the lead truck and then leading three to six trucks. The following trucks could have mostly off duty and lower paid drivers. There could be switching of drivers like changing shifts to the lead truck. It would enable close to 24 hour operation.
So then when would things shift over to fully automated ? Commercial airplanes basically operate in automated flight and landing but still have pilots.
There needs to be definitions around what is fully automated.
I would define fully automated as meaning that the car can operate in an fully automated mode and does not have to have the manual mode eliminated.
The Oxford University’s Mobile Robotics Group (MRG) RobotCar has a modified Nissan LEAF. Lasers and cameras are subtly mounted around the vehicle and taking up some of the boot space is a computer which performs all the calculations necessary to plan, control speed and avoid obstacles.
The sensors and computers build up a three-dimensional map of the route. This is augmented by semantic information such as the location and type of road markings, traffic signs, traffic lights and lane information, as well as aerial images. Since such things can change, the system can also access the internet for updates. Only when the system has enough data and has been trained enough will it offer to drive the car. They have modified the base Nissan LEAF systems to allow complete fly-by-wire control. Everything from the steering to the indicators can be manipulated by the main vehicle computer in the boot. RobotCar senses the world in two main ways. The first uses a pair of stereo cameras to assess the road and navigate, much like a human driver’s eyes. The second is a little different and uses several lasers mounted around the vehicle. These sensors assess the 3D structure of world and also improve performance at night.
The MRG team sees an immediate future in production cars modified for autonomous driving only part of the time on frequently driven routes.
I would define a modified car the learns your frequent routes and can take over driving your commute as a fully automated car.
If I am in the drivers seat but no longer have to steer and step on the accelerator or brakes and a computer or robotic system is safely performing those tasks for at least one complete trip then that is a fully automated car. It also has to be legal for commercial purchase and licensed for general public operation.
I would consider a car or truck that can enter platoon operation (without manual operation) to automatically follow another car or truck as a fully automated car.
A bus or taxi or vehicle that operates to carry passengers and is not on rails (or some other physical guide) but operates with robotic control without an active driver is a fully automated car. Even if it repeats a route that like at Heathrow airport.
It has to be able to get up to 25 mph for operation on a common standard road. I would not count the tow tractor bots which have a top speed of 2.4 mph.
The Robotic Industrial Mining trucks count
Starting in 2008, autonomous trucks have navigated in the complex mining environment and can haul a 290-tonne (320-U.S.-ton) payload of overburden and ore without a driver. At the West Angelas Mine, the trucks are operated and controlled entirely using a supervisory computer at an operations center. In the future, the trucks will be controlled 24 hours a day from a remote operations center located more than 1,000 km (621 mi) away in Perth.
If no one is inside the vehicle driving then it is a driverless vehicle so long as the remote human operator is not actively steering 100% of the time. There are a few tens of such vehicles now and there will be hundreds in a few years. This is detailed in the Rio Tinto mine of the future plan. Another part of the plan is driverless trains.
I think there will be several robotic trucks that will drive coal and mine material from mines to plants on private roads.
The future where the millions of semi-trucks shift over to mainly control center operation where one person is overseeing dozens or hundreds trucks would be a future with fully automated driverless trucks.
There are commercial driverless trucks now but they are mostly not on commercial roads but are licensed for private operation.
When will Driverless cars or trucks arrive and when will they be common ? They arrived in 2008 but were still classified as an experiment until 2011.
They reduced the staff needed for operation. Human jobs were lost / displaced.
There will be hundreds of such vehicles by 2017.
There will be over one thousand by 2020.
I think the Google cars will be operating on a more general basis by 2020 under private (non-Google ownership, purchased by a company or person that did not build them.)
There will be platooned robotic vehicles and control center vehicles by 2020 for private (non-experimental) operation.
I think there will be over 1 million robotic cars or trucks by 2030.