IBM recently gave journalists a sneak peek at some of the revolutionary computer technologies that scientists in its eight research centres worldwide are developing. Alma J Buelva reportsWith a $5 billion budget, IBM Research today is inventing technologies that aim to smooth the interaction between man and machine, allow computers to be worn, and help make pervasive computing devices that automate many of today's mundane tasks.
With the theme "The New Blue - Technology for Life", IBM took journalists on a tour of the future the way IBM sees it. In particular, the IBM Research team demonstrated advances in ease of use, pervasive computing, advanced computer-human interaction, and information management. The preview included wearable PCs, new search algorithms, as well as speech and handwriting recognition.
Care to include a PC in your everyday garb? IBM Japan Research Center may have just the thing you need. Using high-density semiconductor and small design technologies, the wearable PC includes an Intel MMX 233MHz motherboard, a battery pack, a compact controller with infrared and track point, and a head set. It functions like a ThinkPad 560x and runs Windows 95 or Windows 98 and other standard software. The tiny LCD monitor actually works like a hologram, and positioned properly, lets the user see the screen as a large display.
The prototype wearable PC comes with an IBM microdrive (1in disk, 5mm thick, 20 grams, 340Mb capacity), a 256KB external L2 cache, 64MB (EDO) RAM, 2MB video RAM, infrared capability, and a USB port. The battery can last from an hour to two. The battery weighs 299 grams while the headset is 50 grams. Total weight is 450 grams, including cables.
Why a wearable PC?
According to Shigeo Mori, the lead mechanical designer of Wearable PC at IBM Japan Research Center's embedded systems business unit, their intention is to produce the next-generation computer using IBM technology and to see what can be done in the future.
"It is ideal for people who travel a lot. It's hands-free, too. We can technically release it next year, but that is also still in discussion," said Shigeo, who was one of the mechanical designers for the ThinkPad 530 and 535.
The first successful prototype was completed five months after IBM started the wearable PC project last May. If it is finally released commercially, Shigeo said, the Wearable PC would probably cost more than a Thinkpad.
To the disappointment of the press, nobody was allowed to try out the Wearable PC, because the prototype was so fragile.
At IBM's Almaden Research Center, a group of experts on Information Management Principles is working on new approaches to sift information from the Web. Known as the CLEVER (Clientside Eigenvector Enhanced Retrieval) search algorithm development team, their goal is to improve the distillation of Internet links compared to the random style of existing search engines today.
Andrew Tomkins, a research staff member at the Almaden Research Center and a member of the CLEVER team, said unlike the traditional Web search, CLEVER will turn many informal opinions into one confident assessment to deliver the best list of sites. Tomkins said the process is an iterative one that repeatedly re-evaluates the judgements (site suggestions) for destination pages. Once a query or search is made, CLEVER will go through three steps to arrive at the list of best sites possible. These, however, will be accomplished in the same amount of time, if not less, than Alta Vista takes to perform a query, said Tomkins.
Unlike traditional searching, CLEVER will: make the best guess about everybody's quality (site suggestion) as a judge; combine all the judgments (suggestions) using this information to generate a first-cut solution to the problem; and use this solution to improve the assessment of the judges.
"Alta Vista and Yahoo may have great judges, but CLEVER has 100 million lousy judges who keep great lists," Tomkins said. Unlike querying, CLEVER also adopts what it calls "trawling" to discover emerging communities even before the designers become aware of them. CLEVER's trawling figures out when two things are alike or when something is a sub-set of another. Trawling, Tomkins explained, is a process of searching an entire snapshot of the Web to find all similar Web structures.
In addition, CLEVER also makes use of taxonomies, which is a process of adding new and best Web links as they become available. "With taxonomies, CLEVER can find for the users great pages about a topic and automatically update these pages. The taxonomies will be automatically maintained by CLEVER," he said.
CLEVER has been on the designing board for about a year and a half now. Tomkins said their group is talking to Internet portal sites about licensing.
IBM itself, however, has no plans yet to make CLEVER a full-blown search engine product.
IBM is also doing extensive and long-term research into speech technology with the aim of making tools that understand dictated commands without prompting by hardware. With emphasis on natural language, the new technology should adjust to the user and not work the other way around.
An early prototype demonstrated here has a multi-modal feature which combines pointing and speaking to complete the dictation task. It is not a full product yet, but it may replace or be an upgrade to IBM's existing ViaVoice Gold speech recognition software. The multi-modal version can run on a 200MHz Pentium MMX machine.
"The ability to express ourselves multi-modally is important as it is akin to the way human beings communicate," said John Vergo, research staff manager for the Advanced Computer-Human Interaction Applications at IBM T.J. Watson Research Center in New York. "The application of this technology will be important in every IBM business and for telephony embedded and call centre applications."
Vergo said speech technology is still imperfect even today, but he said a certain degree of imperfection is normal because even people can't always comprehend everything correctly. IBM's plan, however, is to make it popular among less savvy users.
Vergo's project is being designed to require only 15 minutes of training from the user. It can get accustomed to different intonations, and will support different languages.
Aside from turning the audio of dictated speech into text, IBM's multi-modal speech software is also being developed to take commands from the user by means of a more extensive collection of phrase commands.
IBM's Speech and Pen Systems group is also on the edge of new solutions co-developed with the pen company, Cross. The result of the partnership is a digital notepad called Crosspad XP that actually makes use of paper.
The portable digital notepad was conceptualised by IBM, with Cross providing the digital pen and the cartridge slot tray for the ink. It comes with a digitiser, small LCD display for status, memory to store data, and a controller with four AAA batteries that last for three months. The Cross pen contains an RF transmitter and uses a AAA battery that lasts for six months of regular use. The system also makes use of a serial connector.
Unlike typing on a keyboard, scribbling on a Crosspad gives a familiar and non-intrusive interface and allows the written document to be archived or stamp dated. The handwritten notes can be uploaded onto a PC as graphics files.
IBM Ink Manager
Toby Maners, Pen Segment manager, said unlike imaging or scanning, the Crosspad uploads only the inking and leaves out other elements that might be present on the medium or paper. Users who also accidentally write over previous text can rely on the IBM Ink Manager software that supports the Crosspad to "lift" the second layer of writing and have it pasted onto a clean page.
Maners said the Crosspad is ideal for use during meetings because it makes no obtrusive noise, unlike typing on notebooks. She added that the product is a big improvement from earlier models first released in 1990, which weren't successful because of the medium, cost and quality issues.
Cross sells the Crosspad in the US, Europe and even on the Web. Maners said they are still discussing with Cross how to market the product worldwide.
IBM experts from different disciplines are today tasked to deliver ease of use. According to Tony Temple, an IBM Fellow and vice president who oversees these efforts, IBM now has 11 multi-disciplinary teams that regularly meet on this topic. To date, Temple's ease-of-use efforts target the continuous development of a natural user interface that is smart, multi-modal and realistic.
This user-centred design looks at how to facilitate the varied tasks that users want to perform on their computers, while considering other factors like reliability and cost.
"Ease of use doesn't come out just from playing with the mouse or keyboard. It involves a lot of technicians' work and culture to achieve an important dimension so computers can assist, for example, in shopping and trading," Temple said.
IBM's ease-of-use efforts are driven primarily by users' expectations, Temple said. IBM approaches this area by measuring the customers' opinion on a product's usability based on their needs. They talk to users for low-level design and development and for high-level external designs. They also evaluate competitors' products. Before beta and benchmark stages, Temple said they also do a design walkthrough with the identified audience for evaluation and validation.
"Computing is about people, not machines," said Temple. "Users' choice usually drives the standards and they are getting upset about poor usability these days and it's driving the whole effort in the industry."
Temple said ease of use is always one of the top five factors that are important to users. There are companies which only buy products that have been tested for usability and with documented reports available to them, he said.
Temple's department is not exactly a business group at IBM. It's an initiative that started in January 1997 under the software group. It is also not the centre of all activities related to improving user interfaces, because each division has its own separate efforts, Temple explained. He acknowledges that there will never be an ideal user interface as it will always be a continuous process subject to different users' needs or preferences.
Cellular phones, pagers, digital cameras, and personal digital assistants are just a few examples of pervasive computing devices that conveniently allow individuals to access information and act on it when and where they please.
Pervasive computing devices
Today, people are getting so addicted to pervasive computing devices that even IBM wants a part of the pie. In hardware, IBM has its microelectronics, microdrive storage and scalable servers. The company also thinks these pervasive computing devices can benefit from its user interface technology, ProxyServer, and other software tools for Java and security.
"IBM's interest is not to produce the devices but in providing the companies that provide these gadgets the technologies they don't have," said Shon Owens Saliga, marketing director for the IBM pervasive computing division. "We're interested in bandwidth and standards that surround the devices that deliver the information."
IBM's other goals in this area, Saliga said, are to ensure that the marketplace is open, and to bring global end-to-end solutions. "The more device manufacturers there are, the more the products will become cheaper. Competition has produced so many great products today," said Saliga.