 |
|
|
The art of miniaturization and the advent of increasingly
lightweight materials notwithstanding,  soldiers
in the field are still required to lug heavy loads on their
back. The Defense Advanced Research Projects Agency (DARPA)
has long been interested in exoskeletons that could help soldiers
carry more weight without extra effort. Now researchers at
the University of California-Berkeley's Robotics and
Human Engineering Laboratory have devised prototypes of a
self-powered exoskeleton that show great promise. The BLEEX
(Berkeley Lower Extremity Exoskeleton) allows a wearer to
carry a 70-pound backpack with ease. BLEEX consists of two
leg braces that strap onto combat boots; the backpack connects
to the braces at hip level. A small engine (one version uses
gasoline as fuel) provides power to the computer and hydraulic
system. A local area network processes data from 40 sensors
and operates hydraulic actuators to keep the device in step
with the wearer. Latest versions allow a BLEEX pilot to walk,
squat, bend, and climb stairs. Eventually, says Homayoon Kazerooni,
the mechanical engineering professor who oversees the lab,
running and jumping will be possible. Though Kazerooni says
a gasoline-run exoskeleton is no more dangerous than a jeep,
he's not satisfied that gasoline should be the fuel
of choice. "We're experimenting with all kinds
of power supplies." Except, of course, the human kind.
—Thomas K. Grose
|
Innovation and Its Discontents:
How Our Broken Patent System Is Endangering Innovation and
Progress, and What to Do About It
Authors: Adam B. Jaffe and Josh Lerner
Publisher:
Princeton University Press
 The
U.S. patent system was designed to protect inventions and
cultivate innovation. But two leading experts— Brandeis
economist Adam B. Jaffe and Harvard investment banking specialist
Josh Lerner—argue in a new book that the system is bogged
down by litigation, thanks to changes made in 1980s. Unless
there are fixes, U.S. innovation could be hampered, they claim
in Innovation and Its Discontents. Many infringement cases
are so complex they're beyond the ken of juries and
judges. They suggest that "special masters" should
preside over the trickier parts of such cases. They also argue
that too many dubious claims are being granted and would like
to see a more robust, multilevel patent review system installed.
Moreover, the authors want a procedure in place that could
allow for a re-examination of patents already issued. —TG
|
Last fall, an American Council on Education Report noted
that minority enrollment at the nation's colleges and
universities soared 122 percent between the academic years
of 1980-81 and 2000-01, rising from just under 2 million to
4.3 million. But is that impressive record in danger? This
year, many top universities report that applications from
African-American students have slumped. They were down 28
percent at Ohio State University, 25 percent at the University
of Michigan, and 12 percent at the University of California-Berkeley.
Several issues may be at play. Affirmative action programs
are still alive at most schools, despite last year's
Supreme Court ruling that said that although race can be a
factor in admissions, it can not be the deciding factor. That
perhaps led to the false impression that affirmative action
programs have ended when they've only been modified.
Another factor: There have been a lot of big tuition hikes,
and many grants for minority students now fall short of students'
needs. Budget cuts have also meant less recruitment in low-income
neighborhoods. Sam Agronow, Berkeley's director of policy,
planning, and analysis, says that headlines last year warning
that the University of California system would be cutting
back on enrollments, coupled with less money for recruitment,
might have discouraged many black high school seniors and
caused a drop in applications. And money available for outreach
programs remains tight at Berkeley, Agronow adds. It's
worrisome, he admits, but it may not portend a trend. "It's
too early to use the T word." —TG
|
 Biometric
security technologies—electronic devices that scan physiological
or behavior traits, such as faces, voices, fingerprints, handprints,
and signatures—should become a $1.2 billion market this
year, the International Biometric Group (IBG) says. That's
up from $719 million last year, a 40 percent leap. The main
buyers now are government agencies, with growth building in
the corporate sector. But the market will really take off
when products for consumers become more available. IBG
expects the market to surge from $1.85 billion in 2005 to
$3.7 billion in 2007, hitting $4.64 billion the following
year. Fingerprint authentication will be the prime mover in
the field; the market for that technology should streak from
$198 million last year to $1.5 billion in 2008. The
market for face-scanning devices should grow from $50 million
to $802 million. So, be prepared for your close up. —TG
|
 The
mechanical engineering department at the California State
Polytechnic Institute knows how to say thank you. They have
developed a delicious chocolate bar in the shape of a gear
for promotional purposes and to thank alumni donors and friends
of the college. The molds were designed in Auto-CAD by mechanical
engineering assistant professor—and chocoholic—Glen
Thorncroft. "They're designed with an involute
shape so the teeth would mesh. The gears had to be correct.
All the ME's out there would have made fun of us if
the gears weren't engineered correctly," says
Thorncroft. — JO ANN TOOLEY
|
Want to talk to someone at the University of Arkansas-Pine
Bluff (UAPB)? Well if you've got a headset/microphone
for your PC, just go to its website—www.uapb.edu—and
use "Click to Talk," an online service that will
let you call anyone on campus from your computer. The school
is a pioneer in converting to Voice Over Internet Protocol,
a phone service that piggybacks on the Internet and sends
voice data digitally to phones or computers. Now other schools
are going to net-based systems, including Dartmouth, Brandeis,
and Brigham Young. And many more, including the universities
of Nebraska and Chicago, are considering making the switch.
Maurice D. Ficklin, UAPB's technical services director,
says he's constantly getting calls from interested schools
such as Harvard, MIT, and Georgetown. IP systems are expensive
to install, but costs can be recouped. Outgoing long-distance
calls are much cheaper. Schools can also reduce maintenance
charges by combining servers for data and voice networks.
UAPB has quadrupled its number of campus lines, but its monthly
fees to a local telecom have dropped from $40,000 to $16,000.
Although it still charges students for phone service, it now
pockets that revenue instead of passing it on to a phone company.
"Cost recovery," Ficklin says, "is the main
reason for going to IP." —TG
|
OSAKA, JAPAN—Eighteen students, ages 22 to 71, have
become Japan's first class of "engineering entrepreneur"
majors, at an unusual program launched last spring at Osaka
Sangyo University (OSU). Japanese engineering schools have
been adopting more real-world and engaging elements such as
project-based learning and internships. The two-year OSU entrepreneur
program marks perhaps the most radical renunciation of conventional
classroom learning in this country. Students are divided into
teams of three, assigned specialist instructors, and shepherded
as they try to convert paper ideas into profitable products.
Classrooms are kept open holidays, weekends, and late nights.
The program manifesto reads like a page from Dale Carnegie,
calling for strong self-reliance, grasping customer needs
and translating these accurately into designs, teamwork, mastering
the basics of patents, protecting intellectual property, and
obtaining venture funding.
The new program is not meant only for turning out entrepreneurs.
It is meant also to bolster the fortunes of the surrounding
Osaka industrial region, famed for its history of craftsmanship
but lately fallen on hard times. The focus will be on low-tech,
high-value-added products, says OSU program faculty member
Osamu Yamada. "IT, nanotech, and biotech require too
much startup capital. When it's make-or-break on thin
capitalization, low-tech is the way to go. Companies in eastern
Osaka have unique know how. It's vital that we partner
with them. We need to sell products that others can't
make. I want our entrepreneur students to firmly grasp this."
In 2000, OSU became the first university in Japan to set
up its own venture business, after Yamada's lab developed
and began marketing a new porous ceramic material with potential
use ranging from medicine to space technology. Plenty of other
ideas are waiting to be hatched in the university lab, OSU
proclaims. —Lucille Craft
|
Like the adage says, there are lies, damned lies, and statistics.
Now a survey by two Spanish statisticians claims that scientific
and medical journals are rife with statistical errors. Emili
Garcia-Berthou and Carles Alcaraz of the University of Girona
reviewed an entire year's worth of Nature articles (2001)
as well as a random sample of 63 British Medical Journal (BMJ)
articles from the same year. They found that at least one
error appeared in 38 percent of the Nature articles and that
25 percent of the BMJ pieces also contained at least one bad
stat. Most of the mistakes were minor—usually numbers
that were improperly rounded off—but it's possible
that 4 percent of them may have inflated the importance of
insignificant findings.
BMJ deputy editor Kamram Abbasi says the findings "didn't
surprise me. We all know that peer review is an imperfect
process." Highly specialized, highly intelligent peer
reviewers rarely catch major research-paper errors, he says.
"So the chances of them picking up on small numerical
errors is highly unlikely." Since few if any findings
are in dispute, and most of the errors are trivial, it's
unlikely the statistical errors were committed intentionally.
The BMJ does have statisticians vet articles, but asking them
or editors to recalculate all the math included in its 7,000
to 8,000 annual submissions would be not be worth the effort,
he says.
In a Web-centric world, however, it is possible to post online
the raw data used for each article printed. Interested readers
could then do their own fact-checking. Both Nature and the
BMJ are considering such a step. But if past experience is
an indication, researchers might not want to cooperate, he
says. "Asking for raw data can be a prolonged, painful
process." Moreover, Abbasi says, a lot of raw data "can
be messy and pretty damn unreliable." Still, he thinks
it's an idea whose time has come, and one that will
lead to more transparency in scientific and medical communications.
Clearly, that's a good thing. —TG
|
 Japan
has hung on to bragging rights for the word's fastest
supercomputer for more than two years now. That's not
too surprising since its top-rated machine—the Earth
Simulator (ES)—debuted at a speed faster than the combined
power of the 20 fastest U.S. supercomputers at that time.
The ES can crunch numbers at a mind-boggling 35.9 teraflops—or
35.9 trillion calculations per second. The latest listing
of the world's top 500 fastest supercomputers—compiled
by Jack Dongarra of the University of Tennessee-Knoxville,
with colleagues from Germany's University of Mannheim
and the Lawrence Berkeley National Laboratory—was released
in June and it showed ES still atop the heap. The new number
two, Thunder, housed at the Lawrence Livermore National Laboratory,
clocks in at 19.94 teraflops. That's impressive, but
no threat to ES.
Nevertheless, the June ranking may be ES's swan song.
IBM and Lawrence Livermore are toiling to finish a new monster,
Blue Gene/L (which will be used for genetic research), that
in theory will reach an astounding 360 teraflops. While no
machine ever hits its theoretical limit, clearly Blue Gene/L
will be dazzlingly fast. Two prototypes of Blue Gene/L made
the top 10, checking in at No. 4 and No. 8 with speeds of
11.7 and 8.65 teraflops, respectively. Dongarra says if IBM
can get the behemoth running by November 1, it will likely
take the top spot on the January list.
The Oak Ridge National Laboratory plans to build a new science
and engineering research center that will house a Cray computer
initially capable of hitting 50 teraflops and, potentially,
250 by 2007. But Dongarra says there are plans to upgrade
ES, so don't consign it to the history books just yet.
The June list also showed the growing prominence of cluster
machines, which harness the power of banks of PCs, providing
supercomputer speeds at less cost (ES, for instance cost about
$400 million to build). Cluster architecture is now the most
common type of supercomputer: there were 291 cluster machines
on the June list, up from 208. "The price point for
clusters," notes Dongarra, "makes them very attractive
for many applications. However, there is still a place for
specifically designed scientific computers." —TG
|
AUSTRALIA—Engineers often have to make complex decisions
that can have serious consequences. Israeli engineer Samuel
Sela, working temporarily in the mechanical engineering department
at Melbourne's Monash University, has come up with a
way to help them navigate the tricky decision-making process.
His method is based on the Pugh Analysis Charts, developed
by Scottish professor Stuart Pugh. Pugh Analysis is a process
for finding solutions by building on the simple strategy of
"pros and cons" lists.
Sela has taken the idea a step further, using quantitative
tools and statistical methods. His method involves assigning
a team and leader to a problem. Team members are drawn from
different relevant areas, such as mechanical, electrical,
and software engineering, as well as management. Team members
first decide which factors require consideration—cost,
performance, and company requirements for example. Each team
member gives the problem a score based on how much he or she
rates each factor. The team leader has the additional responsibility
of rating each factor based on its importance to the company
but doesn't let team members know what those rankings
are. This scoring is then statistically analyzed and the best
alternative presented. "One of the advantages of this
method is that it is ‘personality free', "
Sela says. "No one person can override the views of
another, as each does his or her own scoring." —Chris
Pritchard
|
|
|
|
|
|
|
|
|
|
