Pouring rain and 13mph wind didn’t dampen the mood inside HAX‘s headquarter in San Francisco’s South of Market neighborhood, dotted with startup offices. Tuesday January 10 was HAX Demo Day, the ninth one the company has organized.
Inside, past a long table with coffee and pastries, about a dozen companies displayed their prototypes, some in good working order, others only in mock-ups. Many of them played it close to their chest, choosing to keep the crucial details private until their crowdfunding campaigns go live.
Standing next to an autonomous green-and-white robot on wheel parked on a patch of artificial grass, BEETL founder Xiong Chang picked up what looked like dog poop with his bare hands. The soft plastic poop was fake; but the robot was real. On its home page, BEETL describes its creation as an “autonomous robot that roams your back yard and cleans up after your furry friend.” In layman’s term, it will pick up your dog’s poops. The price hasn’t been set, but Chang says he plans to keep it similar to the cost of a Roomba.
A few tables away, Vitali Wear CEO Cindy Gu was explaining how her bra works — the one on display on a manikin. The company’s Smart Bras have small built-in sensors that keep track of your heart rate variations, postures, and breathing, among others. Those attributes can be used to determine your stress level.
— Digital Engineering (@DEeditor) January 10, 2017
The Fitbit-like wristwatch from Feel measures similar attributes, but for a different purpose. It uses readings like heart rate variation and skin temperature to understand your emotions, such as sadness and happiness. Explaining the science behind the product, Feel writes, “emotion … is characterized primarily by psychophysiological expressions, biological reactions, and mental states such as sweating, rapid breathing and fainting … Electrodermal Response (EDR), which we measure through a Galvanic Skin Response (GSR) sensor, has been proven to be a powerful emotion-related indicator.”
A few feet away from Vitali Wear’s bras and Feel’s wristbands, Elephant Robotics founder Joey Song demonstrates how his robotic arm can “see,” in a manner of speaking. Using a mix of computer vision and machine learning, you can teach Song’s factory-deployable robotic arm to recognize and pick up objects.
A hardware accelerator with one foot in Shenzhen, China, and another in the Bay Area, HAX is a logical host for this type of gathering, partly a show-and-tell, partly a matchmaking event for inventors and investors. At its core, HAX is an investment company that seeks out promising device or product makers and nurtures them through their launch to production. In its own words, “There is no fee, no middleman, no trick to force you to work with someone. Our team will work day in and day out to help you overcome the many challenges ahead.” The company promises to provide “end to end support and a global footprint.”
, HAX’s founder and managing director, offers some insights into the crowded IoT space. “Can you develop a product and ship it? But more important than that, do you have a plan? Are you self-sufficient? Is your product sustainable?,” he asks, as a way for companies to do a reality check. “We’ve already seen on Kickstarter that some companies launched first, and then scrambled to figure out their plans. HAX was founded to counter that. You create the right product, the right strategy, make sure the entrepreneurs joining you have a good understanding of the market … The companies that usually succeed are those with a very long-term view.
Most companies participating in the HAX Demo Day are gearing up for their crowdfunding campaigns. That includes BEETL, Vitali Wear, and Elephant Robotics. There are good reasons for innovative IoT startups to turn to crowdfunding. “These are usually niche products. They’re not like TVs that will sell hundreds of thousands of units,” says“IoT companies are not traditional hardware companies. They need to add software, develop a set of services based on their product. Large companies are not equipped to spend lots of money to bring an unproven product to market.”
There’s something else these companies share — their reliance on machine learning or AI.
Translating several months’ worth of time-stamped heart rates and skin temperatures into stress and emotion indicators is beyond a normal human’s analytical capacity. Keep in mind, in the way Vitali Wear and Feel would apply their technology, this has to be done not just to identify the stress and emotion signatures found in average humans but in individual users.
Similarly, it would take far too much time and talent to program machines to differentiate dog poop from clumps of clay, or nuts and bolts from brackets. On the other hand, a machine learning or AI program can accomplish these in a reasonable time by crunching through a large volume of sample data.
Design engineers are accustomed to working with geometry, with well-defined parameters like stress, heat, and pressure loads. If the products seen at HAX Demo Day are a good cross-section of the IoT space, engineers will have to wrestle with a new challenge — translating sensor-measured values (like heart rates and body temperatures) into the abstract (like stress and emotion).