Write Steady
For people with diseases like cerebral palsy or Parkinson’s, manipulating handheld computers can be tricky. Even if they manage to hold the matchstick-thin styluses and use them to form letters and numbers, the handwriting-recognition software can still translate their shaky strokes into typos. A new text-entry method called EdgeWrite could ease those frustrations. Developed by researchers at Carnegie Mellon University, the system lets the user create each letter or number by following the edges and diagonals of a square hole in a plastic template clamped over the handheld’s text input area. The edges provide stability, and unlike other input systems, such as PalmSource’s Graffiti, EdgeWrite does not depend on the precise path of the stylus. Instead, its software recognizes a character by the sequence of corners hit; it can even be adapted for use with joysticks, touch pads, or trackballs. EdgeWrite co-inventor Jacob Wobbrock, a PhD candidate in Carnegie’s Human-Computer Interaction Institute, is currently providing the software and homemade plastic templates for free via his website; he hopes to find a commercial partner to bring the technology to a wider market.
Executive Squirrel
Overwhelmed with phone calls and can’t afford a secretary? Try a squirrel. MIT Media Laboratory grad student Stefan Marti has built a Bluetooth-enabled animatronic rodent that can manage your calls for you. Like a good assistant, the device gauges how important a caller is and how busy you are before it decides whether to bother you or take a message. Marti says that telecom companies are interested in the critter.
Spoken-Word Search
How do you find one specific song on an MP3 player that holds thousands? You might try scrolling through menus or using a tiny keyboard to type in search terms – but researchers at Mitsubishi Electric Research Laboratories (MERL) in Cambridge, MA, have a better idea: use your voice instead. A Mitsubishi team led by Peter Wolf has developed a voice recognition algorithm called SpokenQuery that lets a user find music simply by saying the name of a song, band, or album–or any combination of the three. Unlike many existing voice-recognition programs, which have set menus and require users to stick to a predefined syntax, SpokenQuery allows the user to put the words in any order and even use partial names. The technology could make it possible to search for not only MP3s but also, for instance, television shows or driving directions simply by saying a few words, says Wolf. The researchers are working to pare down the algorithm’s memory requirements so it can run on many different devices.
Intergalactic Black Box
A data recorder recovered by NASA investigators after the disintegration of the space shuttle Columbia in February 2003 helped them reconstruct the causes of the disaster. But luck played a big part: the device had not been designed to survive breakup or impact. Now engineers at the Aerospace Corporation in El Segundo, CA, are testing a device that can record factors such as temperature, acceleration, and mechanical stresses on a space vehicle as it begins to break up, then detach and carry the data safely through the plasma blaze of reentry. The recorder is about 25 centimeters across and resembles a blunt-tipped rocket cone. Its shield of insulating foam is extremely light, says William Ailor, the Aerospace engineer leading the development of the device. Once it drops into the upper atmosphere, the recorder simply falls to the earth, transmitting its data to satellites before it’s destroyed on impact. Ailor says the company has successfully dropped prototypes of the device from balloons and will have a model ready to fly on expendable rockets next year.
Gauntlet Gab
Using hand gestures to communicate instructions to troops on the battlefield may seem as antiquated as arm signaling on the highway, but it’s reliable and convenient and therefore remains an integral part of troop interaction. RallyPoint in Cambridge, MA, has given the practice a high-tech update in the form of a computerized glove that reads a soldier’s hand signals and relays them wirelessly to troops and commanding officers who may be out of the line of sight. The glove incorporates various sensors that measure how fingers bend and touch and detect the direction and speed of hand movements.
A microprocessor translates the sensor readings into commands – “fall back,” for instance – which can then be sent to other soldiers over radio equipment and conveyed as symbols on helmet-mounted view screens or as verbal commands via an earpiece. RallyPoint is waiting to hear if it will receive its next round of funding for the project from the army.