Fading out? Cambridge's own Xerox PARC

Xerox PARC is the Palo Alto lab famous for developing the graphical user interface, Ethernet, the mouse, and the laser printer. But while it was a petri dish for great research (and great researchers), Xerox PARC famously never did much for Xerox's product line or bottom line.

The Mitsubishi Electric Research Lab in Cambridge is one of the closest things our area has had to a Xerox PARC - a place where cool projects were cultivated that never seemed to have much impact on the parent company back in Japan.

Now, according to this piece in Xconomy, the lab is shrinking. And longtime leader Joe Marks has decamped for Disney, where he's heading up software R&D for the animators.

I'm reprinting, below, an excerpt from an article I wrote for the Globe in 2002, which explored some of MERL's coolest projects at the time.

@ @ @ @ @

"Building Ideas" - Boston Globe, September 18, 2002

- Scott Kirsner

Nervous young researchers are scrambling in and out of a windowless room at the Mitsubishi Electric Research Lab in Cambridge. They’ve just tripped a circuit breaker by trying to run four digital projectors simultaneously.

Once the power returns, a program they’ve written begins to neatly stitch together the images from each of the projectors, creating a massive single image – an IMAX movie for the corporate conference room. But the program keeps crashing in mid-stitch.

If it worked perfectly the first time, it just wouldn’t be the technological bleeding edge. A few minutes later, the snafu has been surmounted, and the projectors begin to harmonize. They paint a rush of colorful imagery against the uneven wall, adjusting flawlessly to its odd angle.

Despite the technology sector’s lingering funk, research groups like the Mitsubishi’s are still working on new hardware and software that could prompt future growth spurts. And Mitsubishi is just one of dozens of commercial and academic research labs in the Boston area that continue to cultivate and prototype promising new technologies.

Even as many tech companies continue to struggle and fail, “the research infrastructure doesn’t die in the Boston area,” says Donald Fraser, currently director of the Photonics Center at Boston University and a former director of Draper Labs, the MIT spin-off that devised the guidance systems for NASA’s Apollo moon missions. “In slow economic periods, you may have a smaller number of people or projects at some of the labs, but you can’t really argue that the sky is falling.”

....

Mitsubishi Electric Research Lab

The researchers at Mitsubishi Electric Research Lab in Kendall Square conduct demos more like magicians than engineers. Hold this telephone to your ear, they’re apt to instruct a visitor. Now pull it away from your ear. Now put it back.

Somehow, you’ve missed not a second of the conversation – everything that the speaker has said while the phone was away from your ear has been stored digitally. The phone senses when it’s against your ear and when it’s not, and when you’re listening after a break, it replays what you missed at a slightly faster-than-normal rate, until you catch up with the real-time conversation. Potentially very useful for cell phones used in the car.

MERL, as it’s known, attempts to improve and simplify the way humans interact with technology. “Everything that Mitsubishi makes, from elevators to power plants to white goods, requires an interface between a human and the machine,” says Joe Marks, MERL’s director. Marks says that despite corporate budget cutting at many Mitsubishi divisions and the recent strengthening of the dollar relative to the Japanese yen, Mitsubishi hasn’t reduced MERL’s budget, and that the lab’s contingent of 45 researchers has remained stable over the past few years.

One team of MERL researchers is devising software so that multiple LCD projectors (which Mitsubishi makes) can be used together, to create large images on sometimes irregular surfaces. Another team is working on analyzing faces in a video image, discerning which direction people are looking, what sex they are, and, perhaps eventually, their age and facial expression. That technology might be used to conduct a demographic analysis of the audience at a rock concert or count the number of people waiting for an elevator on a particular floor to determine how many cars are needed there.

The most fun project: an interactive table called DiamondTouch, which allows several users sitting around it to use their fingertips to manipulate images – like maps or photographs – projected onto the surface. Your fingers become the mouse, and the table can tell the difference between your and those of a colleague sitting across the table. It feels like an entirely new way of working with information, which, the MERL magicians say, is exactly the point.