Tech Talk iconTech Talk

Teaching Cars To Feel Your Pain

Now we drive our cars; in the future, they'll drive us. In between, though, the division of labor will get a little murky, and there could be an unfortunate failure to communicate. Our cars will just have to learn to read us better.

The car that can see a driver's emotional state written on his face could anticipate a bout of road rage and head it off. It could offer advice, or it could just humor the poor, carbon-based life-form. "Yes, that fellow in the red sports car really was rather thoughtless," the car might say, sympathetically. "They really shouldn't let people—I mean, people like himon the road, should they?"

The facial-recognition part of that scheme is under investigation at the Signals Processing Laboratory of the École Polytechique Fédéral, in Lausanne, Switzerland. Researchers trained the system on photos to identify anger or the closely related emotion disgust, then validated the system by testing it on videos, including many taken inside a moving car (provided by the French auto maker Peugeot Citroën, a collaborator in this research).

Read More

California's Dreamin' About Autonomous Car Regulations

Tuesday's public hearing on the regulation of robocars shows that California's Department of Motor Vehicles is sticking to a state mandate to produce a law by year's end. That would make it the first state to accommodate the routine use of autonomous vehicles. (Nevada was the first to approve their use for research purposes.)

The regulators' job is harder than it looks. "The life of the law has not been logic; it has been experience," argued jurist Oliver Wendell Holmes, Jr., long ago. California's regulators must rely on logic alone.

Read More

NASA Needs Coders to Help Prevent Asteroid Armageddon

Savvy coders can soon help NASA defend Earth against asteroid threats and win some cash prizes in the bargain. The U.S. space agency has joined forces with an asteroid-mining company to recruit programmers who can help identify asteroids in the slew of images taken by ground-based telescopes.

Read More

Flock of 104 Spacecraft Set for Launch

(Update, 16 May: After a successful launch in mid-April, KickSat suffered a fault a few weeks later that reset its master clock. The spacecraft reentered Earth's atmosphere on 13 May before it had a chance to deploy the Sprites.)

(Update, 14 March: The launch of the SpaceX rocket that will carry KickSat to space has been postponed until 30 March at the earliest.)

In a few weeks time, a fleet of 104 spacecraft the size of small cheese crackers should be released into space.

This will be the first free-flying test of what could be a new approach to space exploration, one that aims to use bare integrated circuits as spacecraft. Since chips are small and easy to mass produce, it might be possible to launch them by the thousands for use as distributed sensors to monitor space weather, measure the ionosphere, or explore other planets. As Cornell University professor and former NASA Chief Technology Officer Mason Peck described in a 2011 feature for IEEE Spectrum, tiny chip-based spacecraft might ultimately be able to fall gently onto Martian plains without a parachute or coast around for weeks in Titan’s thick atmosphere.

For this mission, each spacecraft in the small armada has a mass of just 5 grams and fits its solar panels, logic, communications gear, and sensors on a single square circuit board measuring 3.5 cm on each side. “They’re way smaller than anything else that’s flown in space—on purpose at least,” says organizer Zac Manchester of Cornell University.

Read More

Electrical Stimulator to Prevent Migraines Receives FDA Approval

Pharmaceuticals are so 20th century. Medicine's buzzy new trend is electroceuticals: ways to treat ailments with electricity rather than chemicals. Yesterday, the U.S. Food and Drug Administration approved a new entry into this field, when it gave the nod to a device that prevents migraines by electrically stimulating nerves in the cranium. The device is manufactured by the Belgian company STX-Med, and is already approved for commercial sales in Europe, Canada, and Australia.  

Cefaly, a battery-powered device that resembles a tiara or a headband, can be used for daily 20-minute sessions. The FDA says this treatment may help migraine patients who can't tolerate or don't respond to medication.

The device works by stimulating the trigeminal nerve, the largest cranial nerve, which carries sensory information from the face to the brain, and which has also been associated with the pain of migraine headaches. According to the Cefaly website, its mechanism is based on the gate control theory of pain. By providing sensory input to the nerves, the device essentially keeps a neural gate closed to pain input. This theory explains why you rub your elbow after banging it—the sensory stimulus overrides the pain stimulus, which is transmitted to the brain on different nerve fibers. 

The FDA's approval is based on several studies conducted in Europe. One study of 67 migraine patients, published last month in Neurology, found that Cefaly users experienced significantly fewer migraines over the course of a month than patients who used a placebo device. However, in a user satisfaction survey, only 54 percent of patients who rented a trial device said they wanted to purchase the device. As those users who returned the Cefaly were shown to have used the device for only about half the recommended time, it seems possible that migraine patients are still more comfortable with popping a pill than wearing a tiara around the house.  

The product video provides more information, although there's a good bit of jargon to deal with. It may help to know that nociceptors are the receptors on neurons that respond to a pain stimulus. 

Image and video: Cefaly

“Naked Truths” About Wearable Electronics

So, maybe you’ve bought a Nike+ Fuelband or a Jawbone UP band. Maybe it’s still on your wrist, and you check its data compulsively. But maybe it’s in a drawer, with your clunky smartwatch, your portable HD radio, and your Microsoft Zune.

Or maybe you are an early adherent of Google Glass, stoically enduring the discomfort and occasional aggression of friends, relatives, passersby, and drunk, enraged bar patrons. On the other hand, perhaps, like app developer Q Manning, you’ve concluded sadly that Google Glass is “great for texting and that’s about it.”

“I don’t think wearable technology has found its niche,” Manning added during a panel session Monday at the SXSW Interactive conference. “We all know we want it, but we don’t know what we want it to do yet. We’re all waiting for someone to solve that problem, but, unfortunately, Steve [Jobs] is gone.”

Manning’s co-panelists certainly had a few ideas, though, and one of them was vividly demonstrated by a pair of statuesque models who took the stage in technologically advanced and yet attractive underwear, prompting a blizzard of camera flashes. The demonstration came as close as could be reasonably expected to fulfilling the promise of the session’s title: “Tech Off Your Clothes: Naked Truths Of Wearables.”

The underwear were prototypes developed in a project called “Fundawear,” explained panelist Jay Morgan of the marketing firm Havas Worldwide. Havas had been hired by the UK-based manufacturer Durex, which was eager to associate its brand more with innovation. Though it offers an extensive line of personal massagers, for example, Durex is generally recognized only as a maker of condoms.

Havas was charged with coming up with a splashy innovation for Durex, and their brainstorming soon centered on the question, “can we do something for [lovers] when they are not together? Can you actually touch someone over the Internet?”

This proposal led to what can only be called a milestone in human civilization: the first ever electrically-engineered underwear. Fundawear was designed by another of the SXSW panelists, Billie Whitehouse, whose firm is called Wearable Experiments. The prototype Fundawear samples are close-fitting black undergarments equipped with tiny haptic electromagnetic vibrators, which produce a momentary sensation much like piezoelectric-based units used to create the quick vibration in a smartphone when a key is touched. The units are strategically installed in the garments—in the female version, in both brassiere and panties—to make contact with sensitive regions of the body. When the wearer’s partner touches his or her smartphone—the sensitive regions are helpfully indicated on a template on the phone’s screen—the wearer feels a gentle frisson, or even a light stroke, depending on whether the partner has touched or swiped the screen.

The engineering challenges were nontrivial, Morgan explained. First, they wanted to create the sensation of a lover’s gentle touch, and finding a technology that could reliably and safely produce that feeling—and be laundered—wasn’t easy. “There was some great work in electroactive polymers at the University of Auckland,” Morgan explained. “We contacted them and they said, ‘where are you going to stick this stuff?’ And we said, ‘down in your pants.’ They said, ‘Oh, that’s probably not a good idea,’” noting that the polymers operate at 4500 volts.

Another challenge was making instantaneous contact, no matter how far apart the partners were. An app pairs the wearer’s smartphone to his or her garment, using WiFi (Bluetooth would be used if Fundawear is put into production, Morgan said). But the users’ phones have to communicate with each other securely and instantaneously. The design team solved that one by using Amazon Web Services, a cloud scheme, to convey the data between phones. Each partner has a secret key to assure security, Morgan added.

Eugenia and Stephen, the models who wore Fundawear at the SXSW session, were delighted with it. “It’s a very light vibration, like a touch,” said Eugenia. “It depends on how you touch,” she said. “If you slide, it’s like a stroke.” The two did not know each other prior to the SXSW assignment, and they each controlled their own undergarment, Eugenia said. “We didn’t want to be too intimate in this situation,” she explained. (Stephen didn’t have a comment.)

Morgan said executives with Durex’s Global Product Development Team are studying the feasibility of putting the wearables into production. He said a decision could be announced in the next several months.

Wearable Experiments’ Whitehouse demonstrated another possible direction for wearables at the session. She was wearing her Navigate Jacket, which pairs to a smartphone app and gives gentle and appropriately timed taps on the left or right shoulder to guide the wearer to a destination.

Asked about the future of the jacket, Whitehouse replied, possibly with tongue slightly in cheek, “The future of this jacket, for me, is inductive-charging coathangers.”

Later in the session, when asked about what people, particularly older people, really want in wearable electronics, she said: “My mother, in particular, just wants a device that repels all other technology. That doesn’t let people track her, or contact her.”

Fighting Buggy Code with Genetic Algorithms

The offspring of a genetic algorithm and a data-structure-reading spider could hunt down and eliminate the software bugs that plague development projects from websites to automotive sensors.

If a 2012 Cambridge University business school study has it right, programmers spend about one-quarter of their working lives debugging. (They devote another 25 percent of their workdays to designing and writing code; paperwork and meetings eat up the rest.)

With a shrug and a sigh, we’ll pass by the obvious implication (kill the meetings and shelve the status reports) and move on to bug hunting, the blood-sport that consumes three months of the work year.

It’s clearly useful to automate software testing, and engineers are doing that whenever they can—particularly in complex applications. But they still have to develop input data sets and testing strategies, and those can be a bottleneck. The programmers can hand-build data sets and test the code unit by unit. They can formulate rules to generate random collections of well-formed data. They can search old files or try to cadge sets someone else once used for something that was sort of similar.

If only you could automate the automation and quickly generate suitable test inputs, and then optimize them to cover as much code and uncover as many flaws as possible.

Software engineers at Germany’s University of Saarland are among the hunters pursuing this quarry. The school’s Software Engineering Department has developed a number of methods for automatically generating tests, particularly for applications with structured inputs. To date, they’ve optimized test-generators for Web pages (a package called WebMate, commercialized by TestFabrik); Java apps (Exsyst); Android apps (DroidMate, whose motto is: “Android apps are great. But when they crash, hang, or violate your privacy, they suck.”); and, most recently, for XML and JSON (XMLMate, “Search-based XML input generation”).

At first glance, the methods seems focused on Web services, but bugs in almost any software project can be fair game.

“The best thing is that we are completely independent from the application area,” says department Chair Andreas Zeller. “With our framework, we are not only able to test computer networks, the processing of datasets, websites or operating systems, but we can also examine software for sensors in cars."

To test a so-called cyberphysical system, like an automotive sensor (photo), “All you need is a way to translate XML into the structured input formats required by the system under test, which usually is a straight-forward translation process,” Zeller says.

The packages are, well, evolving. The most recent test generators apply genetic algorithms to initial seed data and continually improve upon it.

And the latest, XMLMate, helps to automate and improve the seeding process. In a paper prepared for this June's International Symposium on Software Testing and Analysis (ISSTA 2014), Zeller, Nikolas Havrikov, and Matthias Hoeschele describe “search-based” input-set generation: the test-generation software combs through XML schemas (the files that define how objects are to be categorized and treated) and sample test-input files gathered from all over. These are processed into a number of input packages of validly formed data.

These packages are the “chromosomes” on which a genetic algorithm forces evolution. XMLMate swaps, recombines, substitutes, and just generally mutates the input-package data—following the rules of the schema, of course. Then it pumps input package after input package through the “subject”—the program being tested. XMLMate tracks the results and scores each input-package’s relative “fitness.” The fittest test inputs activate more lines of the evaluation program’s code…and provoke the greatest number of errors. The unfit data sets—the dull ones that don’t stray off the beaten path and don’t cause any trouble—lose the evolutionary struggle and are flushed from the gene pool. The trouble-making elite are saved for another round of evolution and reaping.

The Saarland researchers tested XMLMate on six XML-processing programs: Rome, a library for processing RSS and Atom syndication feeds; JEuclid, a library for rendering and converting MathML;  Freedots, which renders MusicXML musical notation into Braille and MIDI audio; Apache Batik and SVG Salamander, libraries for modifying, converting, and rendering Scalable Vector Graphics; SVG Image, a “simple” Java SVG rendering app; and FlyingSaucer, an XHTML-and-SVG rendering library

At this point, automated approaches haven’t achieved complete coverage of every line of the subject’s code. Indeed, coverage ranged from a few percent to just under half. The tests did find, however, that search-based test generation consistently increases coverage by anywhere from 10 percent to (in the extreme case of Freedots) more than 900 percent over randomly generated test inputs. Or, as the paper puts it, “Evolving sample inputs achieve coverage that would not be reached by random seeds.”

Oddly, adding the genetic algorithm to the mix did not uniformly raise coverage per se. What evolving code did clearly do was increase the number of exception errors it could provoke (remember, exception generation factored into the fitness function). Overall, non-evolved, random-test-input packages generated 12 unique program exceptions in the six trial programs; tests that evolved from search-generated seeds turned up 28 errors. (Or, to put it another way, 16 fewer bugs for customers to find.)

Interestingly, the test program also turned up problems in programs that were not in the evaluation group:  They found “one small file” that shocked Firefox into consuming all available main memory and shutting down. Another input set crashed Opera. And some JEuclid test inputs crashed the Java 1.6 virtual machine on both Linux and Windows.

Right now, Zeller and his colleagues plan to make XMLMate available as open-source software…though it is also possible that TestFabrik, their WebMate partner, might commercialize the utility.

Illustration: Randi Klett; Images: iStockphoto
Photo: Oliver Dietze/Universität des Saarlandes

Putting Electronics in People

A baby born five to 10 years from now in a developed country may get a tattoo not long after her first feeding. It would be an integrated circuit, a discreet and flexible affair, smaller than a postage stamp and probably placed on the chest. It would monitor such biometric parameters as electrocardiogram (EKG), physical activity, nutritional status, sleep duration, breathing rate, body temperature, and hydration. By the time the child is two years old, she will have generated and stored in the cloud more biometric data than has anyone alive today, says Leslie Saxon, chief of the division of cardiovascular medicine at the University of Southern California's Keck School of Medicine.

Read More

How Google Glass Can Improve ATM Banking Security

Google Glass could join forces with QR codes to make ATM banking safer even for people who use "1234" as their favorite password. The idea from German researchers could help thwart at least some cash machine skimming scams by turning PIN codes into one-time passwords visible only to individual Google Glass wearers.

Read More

Finding Earthlike Exoplanets

There are several hundred billion stars in our Milky Way galaxy, and probably even more planets. And among those planets, a world named Kepler 62f may be really, really special. It just might have complex life, and the whole cockeyed caravan that goes with it: buildings, music, poetry, love, conflict, cuisine, games, engineering, science, taxes, and partisan politics. It may be decades or longer before we know for sure, but so far, astrophysicists can say that Kepler is a smallish rocky world like ours, and it is "right smack in the middle of the habitable zone," in the words of Natalie Batalha, the Kepler Mission Scientist at NASA's Ames Research Center.

Discovered by researchers operating the Kepler Space Telescope, the planet is 1200 light years away. Without knowing whether the planet has water or any of the other factors necessary for life as we know it, Batalha and her colleagues don't now claim to have discovered a habitable planet. Nevertheless, finding Kepler 62f has reenergized planetary scientists. “This is a dream, finding evidence of life beyond Earth,” Batalha declared. “But it's not a pipe dream. It might not happen in our lifetime, but it could happen in the lives of our children or grandchildren, and that's an astounding thing.” NASA has a “roadmap” of three planned and proposed space telescopes out to the year 2022. These will bring greater technological resources to bear, for example enabling scientists to detect the composition of the atmosphere of planets in distant solar systems.

Batalha spoke at the South By Southwest conference as part of a panel session Sunday titled “First Signs: Finding Life On Other Planets.” She disclosed that the Kepler telescope, which was set into orbit in March 2009, has found more than 3800 planets orbiting fewer than 3000 stars. It has been studying a triangle-shaped patch of the sky between the constellations Cygnus and Lyra, within which it can see about 150 000 stars.

Kepler's mission is to detect potentially habitable planets, so all of the 3800 planets it has detected are at least 85 percent smaller than Neptune. Planets much larger than Earth are considered unsuitable, because they would lack rocky surfaces and be shrouded in hydrogen and helium. Planets much smaller than Earth are also ruled out, because they would lack atmospheres.

Of the 3800 planets found, 25 to 30 are “potentially habitable worlds,” Batalha said. Of those, 62f seems the most promising, being just 40 percent larger than Earth. The closest of those potentially habitable planets is just 15 light years from Earth. It's a neighbor in galactic terms: if you shrank the galaxy to the size of the continental United States, 15 light years would be like the one-kilometer stroll across Golden Gate Park.

“Kepler has set the stage,” said Amber Straughn, another panelist, and an astrophysicist at NASA's Goddard Space Flight Center. “We now know not only that other planets are out there, but that they are common.”

NASA plans to launch two more exoplanet hunters in the next several years. The second of those, the James Webb Space Telescope, is expected to usher in a new era in exoplanetary studies.  “We're going to move from detecting planets to characterizing planets,” Straughn explained. Kepler uses sophisticated detectors to sense the tiny dropoff in light as a small planet moves across the disc of a faraway star. By measuring that dropoff and its duration, scientists can estimate the planet's size and orbital speed. But they can't tell anything about the planet's atmosphere and composition.

The James Webb telescope, on the other hand, will be equipped with spectroscopic instruments that will make measurements of the atmospheres of the planets. The instruments will characterize the light of the star, and compare it with the light streaming through the atmosphere of the planet. Those comparisons will reveal what wavelengths were absorbed in the planet's atmosphere, thereby giving the researchers a good idea of the compounds in it. The difficulty is the thinness of the atmosphere—it can be thought of as a film clinging to the far-off world. “I think that we are at a unique point in history, to build the technologies to move forward,” Straughn said. “Building these awesome telescopes that will let us find that blue marble out there. And the James Webb Space Telescope is our next stop on that journey.”

The telescope project is a partnership among NASA and the European and Canadian Space Agencies. It is scheduled for launch in 2018, will have a 6.5-meter-diameter main mirror. It is to be put into an orbit around the second Lagrangian Point. This position, known as L2, is about 1.5 million kilometers “behind” the Earth, as seen from the Sun. In this position, the craft would orbit the Sun at about the same rate as the Earth. That means that the spacecraft will be able to use shielding to block the light from the Sun, Earth, and Moon, which will all be near each other continuously from the craft's vantage point. Such shielding is crucial because the telescope's hypersensitive instruments will need to be cooled to prevent spurious noise from overwhelming weak signals from distant worlds.

Said panel moderator Alberto Conti, of Northrop Grumman: “When I was in grad school, when we talked about exoplanets...we didn't know if they existed. Now, we know they're out there. You realize it's an incredibly beautiful universe out there.”

Advertisement

Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Advertisement
Load More