news summary (24)

IBM’s ‘Rodent Brain’ Chip Could Make Our
Phones Hyper-Smart
Click to Open Overlay GalleryAt a lab near San Jose, IBM has built the digital equivalent of a rodent
brain---roughly speaking. It spans 48 of the company's experimental TrueNorth chips, a new breed of
processor that mimics the brain's biological building blocks. IBM
Dharmendra Modha walks me to the front of the room so I can see it up close. About the size of
a bathroom medicine cabinet, it rests on a table against the wall, and thanks to the translucent
plastic on the outside, I can see the computer chips and the circuit boards and the multi-colored
lights on the inside. It looks like a prop from a ’70s sci-fi movie, but Modha describes it
differently. “You’re looking at a small rodent,” he says.
He means the brain of a small rodent—or, at least, the digital equivalent. The chips on the inside
are designed to behave like neurons—the basic building blocks of biological brains. Modha says
the system in front of us spans 48 million of these artificial nerve cells, roughly the number
of neurons packed into the head of a rodent.
Modha oversees the cognitive computing group at IBM, the company that created these
“neuromorphic” chips. For the first time, he and his team are sharing their unusual creations with
the outside world, running a three-week “boot camp” for academics and government
researchers at an IBM R&D lab on the far side of Silicon Valley. *** we had several of our
colleagues attend this boot camp and will provide an update in a future QuEST meeting **
Plugging their laptops into the digital rodent brain at the front of the room, this eclectic group of
computer scientists is exploring the particulars of IBM’s architecture and beginning to build
software for the chip dubbed TrueNorth.
'We want to get as close to the brain as possible while maintaining flexibility.' Dharmendra
Modha, IBM
Some researchers who got their hands on the chip at an engineering workshop in Colorado the
previous month have already fashioned software that can identify images, recognize spoken
words, and understand natural language. Basically, they’re using the chip to run “deep learning”
algorithms, the same algorithms that drive the internet’s latest AI services, including the face
recognition on Facebook and the instant language translation on Microsoft’s Skype. But the
promise is that IBM’s chip can run these algorithms in smaller spaces with considerably less
electrical power, letting us shoehorn more AI onto phones and other tiny devices, including
hearing aids and, well, wristwatches.
“What does a neuro-synaptic architecture give us? It lets us do things like image classification
at a very, very low power consumption,” says Brian Van Essen, a computer scientist at the
Lawrence Livermore National Laboratory who’s exploring how deep learning could be applied
to national security. “It lets us tackle new problems in new environments.”
The TrueNorth is part of a widespread movement to refine the hardware that drives deep learning
and other AI services. Companies like Google and Facebook and Microsoft are now running
their algorithms on machines backed with GPUs (chips originally built to render computer
graphics), and they’re moving towards FPGAs (chips you can program for particular tasks). For
Peter Diehl, a PhD student in the cortical computation group at ETH Zurich and University
Zurich, TrueNorth outperforms GPUs and FPGAs in certain situations because it consumes so
little power.
Startup Aims to Beat Google to Market with
Self-Driving Golf Cart
The startup Auro says its self-driving golf cart will lead to autonomous shuttles for theme parks, vacation
resorts, and retirement communities.
By Tom Simonite on August 20, 2015
Why It Matters
Even if limited to private roads, autonomous vehicles could significantly improve many people’s quality of
This golf cart has been modified with sensors and other equipment so that it can drive itself.
Spend enough time on the roads of Mountain View, California, and you might spot one of Google’s prototype
self-driving cars. Visit the campus of Santa Clara University a few miles away, and you can see a self-driving
golf cart that Nalin Gupta says will shake up everyday transport sooner.
Google and automakers pursuing autonomous vehicles are bent on seeing them take to public roads (see
“Proceed With Caution Towards the Self-Driving Car”). Gupta’s company Auro Robotics is focused on the
more modest goal of seeing slower, less showy autonomous vehicles ferry people around the private grounds
of universities, retirement communities, and resorts.
“We are closer to deploying our shuttles in the market,” says Gupta. “It’s technologically much easier.” Like
Google’s cars, Auro’s vehicles require a detailed 3-D map of the environment where they operate. Collecting
that data for a private campus and keeping it up-to-date is easier, says Gupta. Such environments are also less
physically complex, have lower speed limits, and present fewer complicated traffic situations, he says.
Organizations such as universities and theme parks are generally free to operate autonomous vehicles on their
property without regulation. Although U.S. states including California and Nevada have passed laws that
enable testing of autonomous cars on public roads, the legal and insurance frameworks needed for such
vehicles to enter general circulation are lacking.
Auro’s current prototypes are golf carts modified with laser scanners, radar, cameras, GPS, computers, and
other components needed to steer themselves. One is already being tested on the grounds of Santa Clara
University. Gupta says he has signed agreements to begin similar tests at other universities, as well as a
retirement community and a resort in the Bay Area later in the year.
Team Designs Robots to Build Things in
Messy, Unpredictable Situations
Researchers have developed simple robots that can build structures with malleable materials such as foam and
By Julia Sklar on August 20, 2015
Why It Matters
Robots are often limited because they can’t handle malleable materials or work in unpredictable environments.
One of Nagpal and Napp’s robots has no top or bottom. It can keep working even after falling and flipping
Researchers at Harvard University and SUNY at Buffalo are designing robots to function outside of ideal,
predictable environments such as warehouses or factories and instead work in places where there may be
unexpected obstructions, and where predictive algorithms can’t be used to plan several thousand steps ahead.
The goal for such “builder bots,” which are designed to handle inconsistent and malleable building materials,
is to be deployed as disaster relief agents.
Radhika Nagpal, professor of computer science at Harvard, and Nils Napp, an assistant professor of computer
science at SUNY at Buffalo and a former post-doctoral fellow in Nagpal’s lab, have designed two robots: one
that deposits expandable, self-hardening foam and another that drags and piles up sandbags.
Robots built for construction can usually handle only discrete materials, such as blocks or bricks. The materials
these new robots build with are useful in a range of real-world environments, but they are highly unpredictable.
The foam can stick to most surfaces and expand to fill holes, but it starts off as a liquid, so it’s impossible to
know exactly how far it’ll run before it hardens; sandbags are frequently used in disaster relief as retaining
walls, but the granules inside sandbags have a tendency to shift around when manipulated.
To combat this unpredictability, Nagpal and Napp’s robots are equipped with an infrared sensor that takes
scans and assesses the environment in between laying down a building material. The scan is integral to making
the bots so adaptable.