20 Aug 2014
Don't just implement initial specifications, ex-Google Glass lead engineer Anurag Gupta tells SPIE conference.
by Ford Burkhart in San Diego
One of the world’s top optics designers would like to turn the design world upside down.
Instead of just meeting arbitrary specifications laid down by company bosses, Anurag Gupta, formerly head of the optical engineering group for Google Glass and now the lead officer for sensor optics at Nest – the smart thermostat company acquired by Google earlier this year for $3.2 billion - says that engineers must discover their specifications in a long, painstaking process.
Otherwise, the end product may be inferior or more costly due to multiple design iterations.
“Those initial specifications are just lines in the sand,” said Gupta, a plenary speaker at the optical engineering conference taking place during this week's SPIE Optics and Photonics 2014 symposium in San Diego. “They are often based on guesswork.”
“The people who are going to buy the products may not be sure what they want,” Gupta said. That’s why design engineers must “ask the people what they want”.
Gupta said he is now applying the process to his new work on sensors for home automation products. As well as Nest’s smart thermostats, those include motion detectors and smoke detectors that can be switched off by simply waving an arm. The steps, he said, emerged from his 12 years at several companies including work on HP’s digital TV lines and at Google. Altogether he has engineered some 60 products.
“Generally optics engineers are told, ‘We want you to design for this.’ Today’s engineer needs to take part in all aspects of engineering the product design. They need to focus on ways to discover the specifications.”
From smart glass, to defense products, to entertainment, to ski googles where information is provided inside the view screen, there’s a different hierarchy of values, like comfort, price, performance, field of view, or resolution.
“With smart glasses, people are not going to walk in public with a device that does not make them feel good and look good, even if it were free. So style then is number-one. Then there’s comfort, and then price.”
Perhaps the assigned product cannot be made, he said, and even if it can be, it may violate what the people at the top of the chain have set as the market requirement, the value proposition.
“Nobody knows what the specific issues will be,” he said. “People need to feel it, live with it, and decide what they like about it.”
“The end of the journey,” he said, “may be something that someone would never have thought of in a million years.”
So how exactly does a new product engineer learn how to do all this?
“That’s the unfortunate part,” Gupta said. “It’s not something on which any university trains you. I don’t know how to solve that. Maybe they need a course, at Arizona perhaps.” (Gupta himself took a PhD at the University of Arizona’s College of Optical Sciences.)
Social skills required
The engineer, he said, should develop strong negotiation skills to work closely with the suppliers and achieve trade-offs. “And stay there for months, building tools if needed. Their revenues are tiny. They cannot afford large risks. They require a lot of hand-holding, and assurances that the engineer will guard their intellectual property. That trust must be built up before they will let you on the floor. That’s the social side of product design.”
Tolerances for error, he added, are unique for each supplier, and must be negotiated applying the ‘6-Sigma’ quality manufacturing principles.
He said the future of product design belongs to automation. “As the tools increase, the labor costs will go down.”
Given cultural and language barriers among suppliers, optical designers will also have to educate everyone in the supply chain. If there’s an element that must be achieved, then you must “raise the vendor capability” early in the game. You may have to invent the metrology tools by which each component is going to be qualified, Gupta said.
You may need to perform a failure mode and effects analysis, he added, to identify the failure mechanisms, their probability and severity of effects.
“You need to involve the reliability team, revisit the designs, and perhaps re-run the entire exercise from the beginning,” he said. And then, finally: “Somewhere there will be a factory running with your product being manufactured.”
About the Author
Fords Burkhart is a writer based in Tucson, Arizona.