Harnessing the Data Exhaust (from vehicles)

It may be another generation until driverless cars are the norm, but in the meantime we need data to train the software, as well as to assist in the development of new infrastructure and regulatory regimes. A vast amount of data is already accumulating from cars now on the road, and it dwarfs what’s gathered in Google and Uber’s autonomous car projects.

Data from Dash Labs sensors in cars creates visualizations like this one, helping planners and others make better decisions.
Data from Dash Labs sensors in cars creates visualizations like this one, helping planners and others make better decisions.

While it may be blasphemy in Silicon Valley, perhaps Marc Andreessen and Elon Musk are wrong – at least when they say that the driverless car is very close to being a mainstream reality. Fully-fledged autonomy is something technologists, car enthusiasts, and safety advocates have fantasized about for decades, and that’s probably why there’s a certain impatient imagination common to those who want the future now. As Fortune’s Erin Griffith wrote about the pursuit of the driverless car in 2016, “Silicon Valley is great at building hype.”

So, why the delay in letting robots ‘take the wheel’? Well, while the sensors may be in place, there are many parts of the jigsaw still left unsolved.

It will be another generation until driverless cars are the norm, based on the experience at  connected car platform Dash, with nearly 400,000 users whose data they analyse for insights (co-author Edis is Dash’s founder). That experience suggests we will need data to train the software, as well as new infrastructure and regulatory regimes. Most importantly, we will need acceptance from a variety of stakeholders, including consumers, carmakers, dealers, mechanics, insurers and many more. Those things aren’t going to be solved tomorrow. It will take a generation before the capability becomes mainstream.

Many of these barriers simply haven’t been figured out yet, despite billion-dollar startup acquisitions by Detroit and the recent glut of state-level driverless car legislation. (New York State in early May announced it would begin to allow driverless car testing.) These laws are a move in the right direction, but it will be years before the federal Department of Transportation and the National Highway Traffic Safety Administration are able to turn the gears to enable the biggest transportation advance of the century. But we do have a chance to make our cars a hell of a lot smarter today. And we can do that with data.

It’s bordering on cliché to call data a valuable currency in the tech economy. But according to a recent McKinsey study, the market for automotive data could be worth $750 billion in the next 20 years. Data is answering questions people are asking right now, rather than abstract issues that relate to some future state. So it’s no surprise that the sheer volume of data from cars currently on the road dwarfs what has accumulated on the odometers of Google and Uber’s autonomous car projects. The hundreds of millions of miles collected by companies like Dash, Otonomo, and Zendrive tell us a lot more about actual driver behavior than any quirky online morality test ever will.

For example, when it turned out that Volkswagen had been falsifying the emissions results of its ostensibly clean diesel fleet thanks to some nefarious software, the auto industry knew they were in for a crisis of consumer confidence. If one of the world’s largest manufacturers could lie to its customers, why wouldn’t the rest of them be doing the same? Dash’s technology was able to glean an early insight into VW gaming the system, before the story even broke. And indeed, this is why Ford approached the company, as a neutral intermediary, tasking it to analyze data generated from its fleet’s onboard diagnostic (OBD-II) port. That standard part has been on every car manufactured since 1996, and acts as a direct connection to a car’s central computer. By analyzing real-time data, Ford and Dash were able to definitively prove the performance of Ford’s fleet vs. its competitors, and in doing so, rebuild trust between industry and consumer.

And that same underlying data set has been used for everything from insurance risk algorithms (and consequently cheaper premiums) to predictive algorithms for component failure that can save manufacturers billions. More unexpectedly, both hedge funds and billboard advertisers have begun working with telematics marketplaces to gain a new information advantage.

So, while data have been called the “new oil,” harnessing this exhaust of data has proven to be ripe for monetization –  sprouting a multi-billion dollar industry. It harvests and analyzes everything from your favorite soda brand to who you might vote for in the next election. The auto industry has been a laggard, not because it’s not interested in how its customers behave, but in part because sexy advances like lane departure warnings and hands-free parallel parking—not to mention Tesla’s Autopilot, one of the first so-called “Level 2” autonomous vehicle functions—sell more cars than anything else. Slowly, car companies are beginning to see the value of data to enhance everything from streamlined maintenance—a key ingredient to retaining customers—to being able to partner with third parties to provide a suite of services that customers need ever after they’ve left the car.

Perhaps most compelling, though, are the potential impacts that driving data can have on safety. Tracking driver behavior can unearth insights for smart city policies, such as New York’s Vision Zero project (a safety initiative by Mayor de Blasio that aims to reduce fatalities on the road). It can also be an effective bridge to the driverless era. “These telematics marketplaces help cities identify problem areas and driver behaviors,” says Sarah Kaufman, assistant director of the Rudin Center for Transportation at NYU. “They equip cities with the data points to prepare for driverless cars to optimize effective mobility and safety.”

Driverless cars aren’t going to be here tomorrow, but that shouldn’t stop us from making the roads smarter one car at a time through analyzing the 3.1 trillion miles driven by American motorists every year, and the trillions more by drivers all over the world. That information can be a constant source of information. It can feed machine-learning algorithms fueling the race towards true driving autonomy, and the backbone for usage-based products from insurance companies jostling for new customers in a crowded marketplace. It can help resource planners and engineers make streets around the world safer for drivers and pedestrians alike.

Related Posts
See All

How Big Data Effects Cities

It may be another generation until driverless cars are the norm, but in the meantime we need data to train the software, as well as to assist in the development of new infrastructure and regulatory...

Creating a Smart and Equitable City

It may be another generation until driverless cars are the norm, but in the meantime we need data to train the software, as well as to assist in the development of new infrastructure and regulatory...

Could Crowdsourcing Make Better Cars?

It may be another generation until driverless cars are the norm, but in the meantime we need data to train the software, as well as to assist in the development of new infrastructure and regulatory...

XR is Having a Moment – Will Headsets Share the Spotlight?

It may be another generation until driverless cars are the norm, but in the meantime we need data to train the software, as well as to assist in the development of new infrastructure and regulatory...