November 6th, 2017 — “The era of the human-driven automobile, its repair facilities, its dealerships, the media surrounding it — all will be gone in 20 years.” So wrote automotive extraordinaire Bob Lutz yesterday for Automotive News as part of its Redesigning the Industry series.
Lutz, who is in his mid-eighties, has a rich history of leading automotive companies, so his perspective is worth listening to. But the grim picture he paints of what the near future holds seems to ignore the unique dynamics of the auto industry.
He’s held this view for awhile. Last November, Lutz said automotive branding essentially was dead. We’re all going to be driven around in pods standardized based on categories ranging from low-end, medium to high-end, according to him.
In his latest piece, Lutz argues government will legislate the end of human driving within 20 years — if not sooner. He predicts there will be a wind-down period of possibly five years for people to get rid of vehicles they own. The tipping point will come when 20% to 30% of all vehicles are fully autonomous which will give authorities enough data to see how much safer self-driving vehicles are than human-driven vehicles.
When looking at the numbers, and other dynamics, such as technological and political challenges, Lutz’s aggressive timeline is decades off — despite what some prognosticators in Silicon Valley along with uninformed journalists caught up in the hype might say.
Let’s break this down.
First, the technology challenges.
The reality is, there still is significant work that needs to be done before true level five autonomy will be ready for the masses. There will be niche applications ready for controlled environments. These are already being tested today, but self-driving vehicles ready for any situation in five years? It is a stretch.
Ford says it plans (hopes might be a better word) to launch a level four autonomous vehicle for commercial use in four years — 2021. In fact, Bryan Salesky, CEO for Argo AI, the company Ford has selected to develop its autonomous software and is investing nearly $1 billion in over the next few years, outlined the challenges in a company blog a couple of weeks ago.
He writes, “We’re still very much in the early days of making self-driving cars a reality. Those who think fully self-driving vehicles will be ubiquitous on city streets months from now or even in a few years are not well connected to the state of the art or committed to the safe deployment of the technology. For those of us who have been working on the technology for a long time, we’re going to tell you the issue is still really hard, as the systems are as complex as ever.”
Today’s “autonomous” cars requires LiDAR sensors, cameras and radar all working together to create a realistic view of the vehicle’s surroundings. LiDAR sensors, while capturing three dimensional images, have to depend on camera technology to provide color and texture — still a challenge in poor weather or lighting conditions. Meanwhile, radar is used to detect the range and velocity of nearby images.
Imagine the computing power that is required to sort through all of the data coming from numerous sensors, to provide a realistic picture of what’s happening around the vehicle. Now consider the costs of implementing the sensor technology with software to create autonomy. The cost of development will have to drop significantly within five years for a business case to be made for full-scale adoption.
Furthermore, autonomous driving is not possible without deep machine learning capabilities. To develop the artificial intelligence required for the machine learning requires millions of repetitive tests resulting from millions of miles of driving.
Another consideration is the amount of data that autonomous cars will generate. Connected cars — without autonomous capabilities — will send about 30 terabytes of data each day to the cloud. IHS Markit estimates there will be 152 million connected cars (globally) by 2020. Do the math — that’s more than 4.5 billion terabytes of vehicle data uploaded to the Internet each day by 2020.
Managing — let alone trying to make sense of the data — will be nearly impossible. In fact, the Internet could collapse from the data created by the Internet of Things, says Steve Jurvetson, a partner in Draper, Fisher, Jurvetson (DFJ), a ventue capital firm in Silicon Valley. Jurvetson, an early investor in xTime, an online service scheduling tool, and in TrueCar, believes the solution is neural networks, or tiny machine learning brains embedded into vehicles and other devices.(Neural Networks and Why You Need to Know About Them).
Another critical piece of managing that data is in its infancy. Blockchain technology is just starting to be developed. Currently, we know of only three automakers that have announced initiatives in studying the technology.
A simple explanation — vehicle sensors (which Jurvetson predicts) will self-determine which data to upload to the blockchain. Companies on the cloud will pay to access specific data points on the blockchain. A blockchain will be an additional structural layer to the Internet that sits between the vehicle and the cloud and will house all of the data being generated. (Toyota’s Vision of How BlockChain will Transform Automotive).
Full development of blockchain technology is still five to 10 years out, though.
Based on Lutz’s time frame, there will be between 50 million to 75 million autonomous vehicles on the road in the U.S. within 10 to 15 years. Today, there are at least 250 million vehicles on U.S. roads — 20% to 30% of that number is 50 to 75 million. For that to happen, just about every vehicle produced six years from now will have to be fully autonomous.
It’s not realistic. The technology for true level five autonomy may not even be ready in six years.
But let’s say technology advances fast enough in the next five or so years to make true autonomous a reality. In Lutz’s world, consumers, banks, lending institutions will have five to 10 years to wipe out hundreds of billions of dollars in capital tied to the 250 million non-self-driving vehicles if the government were to mandate the end of private ownership — or human drivers.
Several experts — even Audi of America’s CEO Scott Keough at the J.D. Power Summit prior to the National Automobile Dealers Assn. convention this year got into the act — compare today’s age to that of the horse and buggy era as cars started hitting the road. They argue the shift will be just as fast and dramatic from today’s vehicles to self-driving as it was moving from horses to cars.
Our answer to that? It was a lot easier and cheaper to turn a horse loose in the wild or sell it to the local slaughterhouse (as distasteful as that sounds today) then it will be to scrap a vehicle that still has hundreds, if not thousands of dollars owed on it.
Look at the numbers — we’re going to have to scrap between 25 million to 50 million vehicles a year. For comparison sake, the Cars Allowance Rebate program (Cash for Clunkers), in 2009, cost about $3 billion to scrap 665,000 fuel-inefficient vehicles over a five-month period.
The saying goes, “Politics be damned!” But, politics often is the one doing the damning. And it will be politics that ultimately damns Lutz’s vision. He predicts the government will legislate the end of human driving within 10 to 15 years because of safety issues.
So legislation that will remove a key element of our freedom, while relegating its control to some corporate entity in 10 to 15 years? Good luck with that. No congressional member in their right mind would vote for such legislation. Want proof? Despite 12,000 homicides a year from gun violence, Congress still has yet to move on significant gun legislation.
And any legislation outlawing human driving would encounter countless legal challenges from state and local interests along with the business and private sectors.
Add to that, the glacial pace at which government moves. We’re still debating ObamaCare — eight years after it was passed. We’re also debating CAFE mandates — years after they were initially directed.
Just in the last week, media reports indicated Trump’s administration may drop the vehicle-to-vehicle communication protocols proposed in December. The protocols are considered a critical piece of early autonomous initiatives. Despite the reporting by the Associated Press, the National Highway Traffic Safety Administration is saying no decisions have been made yet regarding V2V communications.
In other headlines, an early draft of the House of Representatives tax reform legislation eliminates the $7,500 federal tax credit for electric vehicle buyers. (The Importance of the Federal Electric Vehicle Tax Credit).
Both are considered critical aspects as we move into the connected vehicle era. Removing them could slow that move significantly.
The point is, nothing our government does is clean or linnear. It’s going to be messy and it will be a fight — a fight that takes years — to legislate the end human driving.
When we step back away from the headlines and the hype surrounding autonomous driving, and look at the numbers and the political and technological challenges, the utopian world of autonomous driving predicted by Lutz and others, in our opinion, is laughably unrealistic.