In Quest for Robotics Startups, Qualcomm Accelerator Scouts for AI
At a time when companies like Google (NASDAQ: GOOG), Facebook (NASDAQ: FB), Baidu, and Microsoft (NASDAQ: MSFT) are engaged in an arms race in artificial intelligence, Qualcomm (NASDAQ: QCOM) is putting a premium on “deep learning” as it solicits startups for its new robotics accelerator.
Qualcomm unveiled its plans to host the intensive, four-month program for robotics startups in October. The wireless technology giant said it would commit over $1 million in aggregate funding for the 10 startups admitted to its accelerator, which is managed by Boulder, CO-based Techstars and scheduled to begin May 25th. (Each company admitted to the program will get a $20,000 investment by Techstars for a 6 percent ownership stake and an optional $100,000 convertible loan from Qualcomm.)
In December, Techstars named former EcoATM vice president Ryan Kuder as managing director of the robotics accelerator. Since then, Kuder and Houman Haghighi of Qualcomm Ventures have been drumming up interest in the accelerator by giving presentations at the Consumer Electronics Show in Las Vegas, and at tech mixers in Boston, New York, San Diego, and San Francisco.
In an interview before their Feb. 9 presentation in San Diego, Haghighi said the team is looking for applications from robotics teams with “core technologies and non-core technologies in hardware, software, and some kind of core intelligence.”
That core intelligence, Haghighi said, should come from some kind of artificial intelligence or deep learning technology. Haghighi, in particular, emphasized the need for robots with the capability to navigate autonomously, using innovations in computer vision and techniques like “supervised learning” and “unsupervised learning” to avoid obstacles. Haghighi talked about the importance of computer vision in navigation, and the different approaches taken by the University of Toronto’s Geoffrey Hinton and Yann LeCun, who left New York University at the end of 2013 to join Facebook.
Is there a conscious effort at Qualcomm to use its new robotics accelerator to drive innovations in artificial intelligence?
When I put the question to Techstars CEO David Cohen in an e-mail, he replied, “I’m not sure I agree that we are trying to drive it to AI specifically. That’s an area of interest, but not exclusively.
“What you are more likely witnessing is that we are more proactively recruiting companies, in addition to looking at who naturally applies, and that this recruiting may have an emphasis on one or more areas.”
Kuder, who was more circumspect than Haghighi, said the accelerator is seeking startups with a broad range of capabilities, including innovative mechanics, intelligent controls, computer vision, sensors, and other technologies. “What’s really critical for us is finding great founders, because great founders can solve problems that were unforeseen,” Kuder said.
Yet even during his presentation, Kuder called out Ping Wang, a familiar figure in San Diego’s tech startup ecosystem who was recently hired to serve as the accelerator’s technical director.
For the past four years, Wang has been running the Ansir Innovation Center, a private tech incubator. But Wang also has a doctorate in computational neuroscience from UC San Diego and studied neural networks and brain-inspired neurocomputing under the pioneering neuroscientist Terry Sejnowski at the Salk Institute.
As the accelerator’s technical director, Wang will review the technical merits of applications submitted by startup teams for the program. (The deadline for applications is March 8, and the program begins May 25.)
Connecting the dots, Qualcomm’s Haghighi explained that Qualcomm’s foray into robotics represents a new market for the kind of wireless innovations that went into the smartphone over the past 10 or 15 years. As former Wired magazine editor Chris Anderson famously declared, “The personal drone is basically the peace dividend of the smartphone wars.”
Anderson, now the founding CEO of drone-maker 3D Robotics, explained that “the components in a smartphone—the sensors, the GPS, the camera, the ARM core processors, the wireless, the memory, the battery—all that stuff, which is being driven by the incredible economies of scale and innovation machines at Apple, Google, and others, is available for a few dollars. They were essentially “unobtainium” 10 years ago.”
As the high end of the smartphone market becomes commoditized, Qualcomm sees robotics as a potentially huge new market for a variety of wireless technologies, including its Snapdragon chipsets, which integrate the core processor, graphics processor, GPS, and cellular modem in one package. The Snapdragon product line is known for greatly easing the design burden on mobile phone manufacturers, and would presumably make it easier for robot design teams as well.
Artificial intelligence, however, is one of the crucial technologies that robots will need to navigate autonomously—and to become useful commercial products. “By coming in really early in this industry,” Kuder said, “we’re hoping we can find those foundational companies.”
By proactively seeking out AI technologies, Qualcomm is playing for the future as it recruits the next generation of robotics startups for its accelerator. At the same time, the wireless giant is looking for ways to seed its technologies in the just-emerging greenfield market for robotics.
Whether there are enough robotics startups ready to make that next-generation leap is a question that should be answered over the next two months, as Techstars notifies the companies admitted to the program.