Driving toward the future
Self-driving companies must prepare for a changing world even if they’re not ready — down to the VR and laser patents
Part of the occasional series “Patented in Pittsburgh”
You pull out your phone to hail an Uber and it arrives, sans driver, just moments later.
A door slides open and you hop in, taking a seat next to your car-pooling comrades who all have funky looking headsets on. That’s when you, too, pull out the goggles and enter a virtual reality.
That’s not actual reality yet — just an idea for which ridehailing company Uber received a patent Aug. 7. It’s titled “Virtual reality experience for a vehicle” in case it decides this is a viable version of the future.
That’s partly how patent strategy works for tech-savvy companies with their eyes on the horizon, explained Chris Storm, legal director of emerging technologies at Uber. You have to prepare for the future even if you aren’t quite ready for it.
“I don’t have to worry that much about, is this a crazy idea?” he said. “Historically, Uber has been good at understanding what customers’ needs are.”
The auto industry has long been adapting to changing technology and that legacy continues: General Motors announced in late November that it would cease production of most compact sedans in the first quarter of 2019 to focus on electrification and automated vehicle technologies.
While there’s no guarantee you’ll be sliding into an autonomous, virtual world on your way to work in 10, 20 or even 50 years, San Francisco-based Uber wants the right to exercise that concept if and when the time comes.
That means securing intellectual property by building a solid patent portfolio. Unlike some major companies that stack up troves of patents to not only protect their own inventions but to sell to smaller firms, companies in the self-driving space generally believe quality, not quantity, is key.
“We want to have a very efficient patent portfolio ... where every asset is achieving some strategic goal,” Mr. Storm said. “That could mean protecting against competitors, sending a message to investors or demonstrating something to the public.”
Sharing secrets
The view that size of the patent portfolio doesn’t necessarily matter is partly a product of close relationships between self-driving technology companies and their suppliers.
Argo AI, for example, would be hard-pressed to build an autonomous software system for Ford without relying on the automaker’s existing patents and some trade secrets relating to its vehicles. Uber, which has an engineering office in the Strip District, would have an issue if it couldn’t see the design specifications of a Volvo SUV, which it is retrofitting with its self-driving system.
At Argo, the Strip Districtbased startup in which Ford invested $1 billion to develop the brains of its future selfdriving fleet, coordination between the software company and automaker is clear, down to the legal team.
Kate Kozlowski serves as vice president and general counsel at Argo AI, where she handles legal and government affairs on a small team. Jason Rogers, who serves as intellectual property counsel for Ford, supports Argo’s patent strategy.
“We do the systems; they do the platform; we work together on other things in between and IP is one of them,” Ms. Kozlowski said.
That collaboration is the secret sauce that the two companies believe will enable autonomous solutions at scale, Mr. Rogers said.
But there are a few factors both must keep in mind.
For example, there are strict confidentiality agreements between employees at both companies.
This is not a new concept, Mr. Rogers said, because automakers have traditionally held similar arrangements with their tier 1 suppliers, companies in the supply chain that provide parts or systems.
Argo intends to find other partners over time, Ms. Kozlowski said, so the arrangement should be replicable. And perhaps more importantly, there should be safeguards in place to keep those supplier-partners from accessing one another’s intellectual property.
“If we brought in another OEM [original equipment manufacturer] customer, we’d have similar confidentiality agreements, as well, and we’d have to firewall that confidentiality,” Mr. Rogers said.
Most companies use a combination of patents and trade secrets to build an overall intellectual property portfolio, he added. But relying on trade secrets is no longer en vogue as younger generations hop from job to job rather than stay with one employer for decades.
A trade secret, according to the U.S. Patent and Trademark Office, is information used in business — like a formula, pattern, compilation, program, device, method, technique or process — that gives a firm an advantage over others who don’t know about it.
Mr. Storm of Uber agrees that patents are becoming more important than trade secrets. He believes that although a piece of self-driving tech may not be detectable today — meaning there’s probably no way that you can figure out an invention even exists, let alone steal the idea — that’s likely to change as the state-of-the-art progresses and companies become more transparent.
“I don’t know how our friends down the street identify objects or do motion planning … so technically you would say, then, that I can keep that all a trade secret,” he explained. “But just because that’s the state of the world today doesn’t mean it’s the state of the world tomorrow.”
Where trade secrets are a part of the equation, however, Argo trains employees not to spill the beans — crucial, given that there is precedent for problems in the emerging self-driving space.
In 2016, a bitter trade secrets lawsuit broke out between Uber and Waymo, wherein a former Waymo employee who moved to Uber was accused of stealing 14,000 confidential files.
So, during calls with potential employees, Argo recruiters make it clear that the company does not want previous employers’ trade secrets. During onboarding, employees are taught best practices to keep from using any secret knowledge learned at a competitor company.
“You have know-how in your brain and you can’t take that out, but you can be mindful,” Ms. Kozlowski said.
Argo even instructs its engineers not to look at competitors’ patent filings to avoid any accusation that they could be stealing ideas.
Not a numbers game
So far, two patents issued to Argo AI are public record, according to documents on the USPTO website.
One is concerned with avalanche photodiodes, a particularly sensitive semiconductor device that converts light to electricity.
In computer vision applications — wherein a machine must find, use and understand images from the real world to “see” — sensors must collect data about incoming light and convert that into an electrical signal that a self-driving car, for example, can understand.
False positives, known as “optical crosstalk,” are created in the process of collecting and translating light data. Argo’s patent seeks to minimize this.
The second patent deals with the light-sensing pixels in a sensor.
In October 2017, Argo acquired a New Jersey-based lidar company called Princeton Lightwave, which specializes in the light detection and ranging sensors that self-driving cars rely on to create a three-dimensional picture of the world.
The applicant name on both Argo patents is Princeton Lightwave, meaning the tech could have been in development for years before Argo was founded.
Still, it’s unclear how many total patents the young company actually has been issued. They do not become public record until 18 months after the earliest filing date, Mr. Rogers said. Argo has been around only since late 2016.
Argo AI declined to release the number of patents it holds until they do become public. Uber also declined. Aptiv and Aurora Innovation, also operating in Pittsburgh, did not comment for this story.
“If you look at other ‘tech’ companies, they have massive portfolios,” Mr. Storm said. “I would venture that most are not serving a strategic purpose ... we do not want to grow into that kind of behemoth.”
Uber uses analytics software to look at large patent portfolios and identify which patents are most important, he said. So if a company has 100,000 patents— using sheer volume to hide key assets — Uber believes it can find the needle in the haystack.
Some companies build out stacked portfolios to not just protect their own inventions, but also to sell the patents to other firms. AT&T, for example, sold 66 ridehailing patents to Uber in January 2017.
That benefits both parties because the larger company has created a revenue stream and the smaller one can get an assist from the research and development budget of a big firm. That helps pad the holes in a startup patent portfolio while on borrowed time.
Still, patent strategy isn’t necessarily about blocking out the competition, Mr. Storm reiterated, or filing tens of thousands of applications. Instead, it’s more about protecting what you have, but being privy to the competition.
“Have you ever heard of the Cold War analogy? All you really want is detente. We don’t really want more missiles facing us than facing them.”