In the past few years, almost every major tech company has become, at least in part, an automotive company. We’re not talking about manufacturing, of course — don’t expect to be seeing any Apple or Google cars roaming the roads, even though Tesla managed to pull that trick off from scratch. Instead, tech companies are working with existing car companies to provide the tech that will power autonomous vehicles — the sensors, the services, the computers, and the networks.
While Apple has been almost bizarrely silent about their plans, every other major tech company has been trying to get some piece of the pie. Microsoft is developing software, Samsung acquired Harman for their automotive equipment and cloud platform, and Google spun off Waymo (which recently established a partnership with Lyft) to develop autonomous driving systems. Uber is working on an autonomous driving system, too, although it might be a bit too similar to Waymo’s.
Right in the middle of all this is Intel, in an increasingly big way. Intel’s in something of a unique position — they can touch on every part of the autonomous car business except for building the cars themselves. They can build processors, create developer platforms for autonomous driving systems, apps, and services, and they even have skin in the sensors game now that they’ve acquired Mobileye.
They’ve even got friends in the automotive industry — BMW and Audi. At CES this January, Intel demoed their autonomous work with the help of BMW, Mobileye, and component maker Delphi. The autonomous cars they put on display during the show are being worked on and tested behind the scenes at driving labs that Intel has set up in Arizona, Oregon, and Germany. And, earlier this month, Intel opened up a new lab in San Jose, right in the heart of Silicon Valley.
So, what’s Intel going to be cooking up in this lab? Increasingly, it’s the little things. The tech needed to make this all work — the sensors, the software, and the computing power — has been around for years. But, it only works in theory. Intel, along with the rest of the industry, now needs to figure out how to make it work in practice — and get it looking like a consumer product.
I spoke with Jill Sciarappo, Intel’s strategic marketing director for their autonomous driving division, about some of those little challenges. While autonomous driving systems don’t actually need to be connected to a network to function, getting them connected to a central network will go a long way in mitigating all those little things that could go wrong.
Sometimes, the stakes can be low — take those awkward crosswalk moments when you’re walking and trying to wave a car through a stop sign ahead of you. Half the time, human drivers will just sit there despite all the hand waving, but just imagine what an untrained autonomous car will be like. Based on how autonomous driving systems have been designed thus far, it’ll take the conservative approach and wait for the pedestrian. Problem is, the walker doesn’t know what the car is thinking, and there’s no driver to communicate with their own facial gestures or hand waving.
That’s why v2p communication is a thing. You might have seen some terms like this before — v2i is vehicle to infrastructure and v2v is vehicle to vehicle. Sciarappo explained that v2p means vehicle to pedestrian — Intel will be using sensors and machine learning (AI, basically) to understand pedestrians and find ways for cars to somehow communicate their intentions.
Other times, stakes can be a lot higher. San Diego is becoming one of the first cities in the United States to build out a connected smart city infrastructure of sensors, and one of the benefits of that is being able to detect where pedestrians are on city streets and sidewalks. That information can be relayed to cars, allowing autonomous vehicles to stop for pedestrians or cyclists before they even come into view. The problem is that that’s just one stream of data out of the insane volume that will be coming from and going into autonomous vehicles. While it’s possible for 4G LTE networks to handle those streams, there will be lag time, and lag time’s not acceptable when dealing with safety features that could be the difference between life and death.
This is where Intel is a little unique. Unlike most other companies in the race, Intel is directly involved in helping to build out 5G networks. As we’ve covered before, 5G won’t just mean faster speeds — it’ll enable low-latency, high-volume transfers and can offer dedicated frequency bands specifically for use with autonomous cars. 4G can make autonomous cars happen, but 5G will enable the safety features that could get the public to trust autonomous cars enough to use them.
All of Intel’s efforts are based around Intel Go, their central autonomous driving platform. Intel and third parties will be able to develop systems and services on top of Intel’s cloud based platform. Instead of having an autonomous driving system run within one vehicle, more value can be added through connectivity — the example cited above, plus other goodies like being able to log in to a personalized account, which would include contacts, files, and preferences.
It sounds amazing in theory, but a lot could go wrong — a big reason why autonomous cars aren’t expected to be on the road and fully autonomous for another five to ten years. If a car’s going to rely on network connectivity for safety features, that network has to be absolutely, 100 percent reliable. Cars will also need to be able to respond to road work, checkpoints, and other unexpected events that sensors and AI might not be able to make sense of right now.
It’s a big risk for Intel. With PC sales slowing and Intel badly missing the boat on mobile in the 4G era (and AMD’s cheaper and increasingly good processors lurking in the rearview mirror), Intel needs a big win. Connected cars represent an enormous new market that could help sustain Intel as the PC market continues to flatline. That could explain why the company paid a king’s ransom to acquire Mobileye — a princely $15.3 billion. It’s safe to say that — along with their new Silicon Valley lab — won’t be the last big bet they make on the autonomous car.