Blog post

The Future of Hardware Design Depends on Machine Learning

By Adam Benzion
The Future of Hardware Design Depends on Machine Learning

In the late 1970s, celebrated father industrial design Dieter Rams became increasingly concerned by the state of the world around him — “an impenetrable confusion of forms, colors, and noises,” he famously said. He then asked himself an important question: is my design good? Rams decided that good design can be measured in a finite set of ten important principles: “Good design is innovative, makes a product useful, is aesthetic, makes a product understandable, unobtrusive, honest, long-lasting, is thorough down to the last detail, environmentally friendly, and last but not the least less important: good design is as little design as possible.”

Rams’ philosophy turned out to be timeless, the stuff of legends. With nearly 50 years passing, he is still as relevant as ever, and it seems like everyone has caught on with his philosophy, starting with Apple, followed by Tesla, Slack, Square, Nest, Dyson, and Amazon. As famed designer Jon Ives once said about Apple’s Rams-inspired design philosophy: “Design simplicity is not the absence of clutter… simplicity is describing the purpose and place of an object and product.”

Today, good design needs something new. Magical experiences.

With 300 billion microcontrollers in the world today and growing, these tiny special purpose computers are running our lives. For example, a microcontroller in a Nest thermostat communicates with live weather data, indoor temperature sensors, your voice, geofencing setting, the HVAC and makes all sorts of “magical” decisions that make your life comfortable, cost-effective and effortless.

Such experiences are integral to the new world of good design. These embedded systems are everywhere, and soon, and progressively in smaller, previously “dumber” and less connected edge devices. Tiny Machine Learning, aka tinyML, is now the fastest-growing field of machine learning technologies and applications, including hardware, algorithms, and software capable of performing on-device vision, audio, biomedical, data analytics more at an extremely low power point measured in mW or even uW. TinyML is becoming so popular that over 100 of the top leading technology companies, from Google to Microsoft, have adopted tinyML as the future wave. You can learn more about this fast-growing technology by joining the tinyML Foundation, which curates quality events – in-person and online – throughout the year, including tinyML Summit, tinyML EMEA, tinyML Asia, tinyML Talks, and tinyML Meetups.

From startups and big tech, everyone thinks that tinyML is the next UX.

There is growing momentum demonstrated by technical progress and ecosystem development. Edge Impulse is one of the pioneering startups working on helping engineers take advantage of tinyML by automating data collection, training, testing, and deployment. Starting with embedded or IoT devices, Edge Impulse is offering developers the tools and guidance to collect data straight from edge devices, build a model that can detect “behavior,” discern right from wrong, noise from the signal, so they can actually make sense of what happens in the real world, across billions of devices, in every place, and everything. By deploying the Edge Impulse model as part of everyone’s firmware, you create the world’s biggest neural network. Effectively, Edge Impulse gives brains to your previously passive devices so you can build a better product with a neural personality.

Another interesting company is our partner Syntiant, building a new processor for deep learning, dramatically different from traditional computing methods. Their Neural Decision Processors operate at efficiency levels that are orders of magnitude higher than any other technology by focusing on memory access and parallel processing. The company claims its processors can make devices approximately 200x more efficient by providing 20x the throughput over current low-power MCU solutions, and subsequently, enabling larger networks at significantly lower power. The result? Voice interfaces allow a far richer and more reliable user experience, otherwise known as “Wow” and “How did it do that?”

Both Syntiant and Edge Impulse have received major industry recognition as Product of the Year and Innovation of the Year at the 2021 tinyML Summit, the premier annual gatherings of senior-level technical experts and decision-makers in the fast-growing global tinyML community.

TinyML code will be everywhere: machine, plant, human, animal.

Good design, as it happens, now has much to do with how you design your embedded sub-system about the software interface and the physical encasement and controls. Today and going forward, good design means good embedded technology. It is expected, desired, and consumed like no other technology globally, and it will impact everything: retail, healthcare, transportation, wellness, agriculture, fitness, and manufacturing.

Designing good products will no longer rely on smooth aluminum surfaces with elegant flat iconography and apps. Good design will mean that machines act as an extension of our brains, feelings, and emotions. Good design will continue to follow Rams’ mantra: Be innovative, useful, aesthetic, understandable, unobtrusive, honest, long-lasting, detailed, environmentally friendly, as little design as possible, and now also an extension of you.

Comments

Subscribe

Are you interested in bringing machine learning intelligence to your devices? We're happy to help.

Subscribe to our newsletter