How cellphones will enhance reality
Despite decades of hype, virtual reality has never really managed to live up to, well, reality. But new applications that blend real and virtual worlds via the medium of a smartphone may have the missing ingredient needed to make the technology take off.
Instead of completely immersing a person in a virtual environment, augmented or mixed reality involves adding digital features to the world they see around them.
Changing the world in this way requires drawing on nearly every faculty the latest smartphones possess – camera, GPS, tilt sensors, digital compass and wireless broadband – to determine exactly what is being looked at, and to find and display relevant extra data.
Hitchhikers' guide ?
So what are the advantages of augmenting your reality? Simply, the technology allows you to point a phone at an object and see an enhanced version of reality on the screen – whether a mountain labelled with its height, a person tagged with their name, or celestial objects properly labelled in the night sky.
An application called Enkin developed by two German researchers for phones using Google's Android operating system provides a straightforward example. It acts like an enhanced satnav system, helping you to find nearby services or points of interest.
Nokia's Mobile Augmented Reality Applications project does a similar thing, drawing its annotations from known points of interest in the satnav software found in Nokia smartphones.
Despite some impressive demos, both these systems are still in development. But another, the Wikitude AR Travel Guide, is already available as an app for Android phones.
By drawing on Wikipedia it can label sites in the phone's view, such as castles, or augment mountain peaks with their altitude, like a kind of universal guide book.
Playing tag
That's powerful, but there is more to come, especially when these apps start to tap into social websites. Wikitude can already tap into location-based contents of Qype, an online user-based review service, allowing people to stroll down a street and read reviews of restaurants, bars or hotels just by pointing the phone's camera at them.
And now Nokia is taking this further by getting your phone to augment people as well as places and objects. Users will be offered a service similar to Google's Latitude software. When the software is running on the phone, icons will appear next to anyone nearby who is also running the program to show if you have mutual friends or common interests.
Nokia researchers are also exploring ways to augment a cellphone user's reality to reveal past actions of others. One spooky example is that you could be alerted with a vibration to the fact that a friend recently passed by – a bit like brushing shoulders with a ghost.
So, rather than stumbling around wearing virtual reality headsets, viewing a digital world through our smartphones takes the technology in a fundamentally more practical direction. By seeing what we see and adding to our knowledge of the world around us, AR-capable phones offer real benefits that are only just starting to be explored.
Winning the arms race over your cyber secrets
Everyone knows they should choose online passwords that are hard to guess: combining numbers and letters, and changing them frequently to make sure they stay obscure. That's good advice, but it won't protect you from a cybercriminal who can simply monitor your keystrokes as you type.
All is not lost, however. Some security researchers think software that recognises how you type, rather than what you type, could prevent fraudulent access to online services – even if the password is compromised.
"Keystroke logging" – covertly recording all keys struck on a keyboard – has been used for years by criminals and law-enforcement agencies alike. It's not just passwords that can be compromised this way; any text typed on a computer can be recorded.
The most common way to do this is to infect computers with keystroke logging programs, typically spread online via viruses and other malicious software – dubbed "malware". But such software can be detected by security programs, and tends to leave a distinct trail when it transmits the data collected, so it's relatively easy, in principle, to protect against.
More worrying are newer ways to log keystrokes that don't require software on the target PC and thus leave no trace.
Open secrets
Last year Martin Vuagnoux and Sylvain Pasini at the Swiss Federal Institutes of Technology in Lausanne, Switzerland, used a wireless antenna to pick up keystrokes from a wide range of commercial keyboards. They claim to have detected and interpreted signals inadvertently leaked by the keyboard to identify keystrokes with 95-per-cent accuracy, and will present their work in full this August.
Laser microphones provide another way to snoop on typists. These surveillance devices can eavesdrop on rooms with closed windows by detecting the tiny vibrations that the sounds create in the window pane.
Earlier this year, researchers from Inverse Path, an IT consultancy firm based in the UK and Italy, announced that they could work out which keys were being typed on a keyboard by training a laser on it. Their system can work through a window, but does require a direct line of sight to the computer.
Personality type
So the security of the standard keyboard, and hence our online accounts, looks to be on increasingly shaky ground. But there is some good news: researchers are now developing software that e-businesses could use to spot an intruder trying to gain access to one of their customer's accounts – even if they type in the correct password.
Just as our handwriting is very personal and can be difficult to forge, so each of us has a unique typing style that's difficult to replicate. For instance, a user who picks out words letter by letter has very different "keystroke dynamics" to a touch typist with a speed of 70 words per minute.
More subtly, some users might tend to capitalise a letter using the left, rather than right, shift key; and others might always have difficulty finding a particular key, leading to a characteristic delay whenever they have to type that letter.
While the idea is not new, it has yet to be widely used for security purposes. This is because there's an obvious drawback: a person's typing behaviour changes with time. Someone might type much more quickly in the morning, when they're fresh, than when they are tired in the evening. Or they might be hesitant when typing a recently changed password.
So if the system learns to identify the user at one time, it might fail to recognise them later when they have used their password a few times and can type it without hesitation – potentially locking them out of their own account.
Tap enhancing
But solutions are on the horizon. In 2005, Sungzoon Cho and Seongseob Hwang at Seoul National University in South Korea suggested a workaround. They asked users to always type their password to a rhythm, such as the beat of a well-known sports chant.
The idea ensures that the user always places pauses in their typing rhythm at the same points in the password (Lecture Notes in Computer Science, DOI: 10.1007/11608288_83). The idea tested well in a trial study, but Cho and Hwang concede that it relies on users possessing a good sense of rhythm.
Felix Racca, CTO at Authenware, a Florida-based keystone dynamics software vendor, suggests another alternative: software that continues to learn and adapt to the user as their typing style changes.
Using this approach, he says, security software could actually become stronger, the longer a particular password remains unchanged. This is likely to appeal to PC users frustrated with the need to update their password every few weeks, he points out.
For now downloading the latest security patches, installing strong antivirus software, and using firewalls will keep you safe from all but the most determined cybercriminal – but we can look forward to some novel ways to let us keep "password1". Damn, did I really type that?
Behind Microsoft's full-body gaming interface
At the E3 2009 gaming conference in Los Angeles, California, this week, Microsoft unveiled a new hands-free, full-body-control system for its Xbox 360 console, codenamed Natal. Using it, players can interact with games simply by talking and moving their body.
Microsoft also claims that it can recognise emotions, and Natal has impressed game players, developers and movie mogul Steven Spielberg alike.
It is undeniably impressive. And you have to wonder: how does Natal actually work? Microsoft is remaining tight-lipped. But we've talked to industry insiders and pulled together some material from the New Scientist archive to suggest how it could live up to the high expectations generated by partially enhanced concept videos like this one.
Camera capture
Although techniques to capture a person's movement using cameras have been in development since the 1970s, doing so without attaching markers to the person's body as Natal does is a more recent possibility.
Organic Motion is one of the few commercial companies currently offering markerless motion capture technology. Andrew Tschesnok, the firm's CEO, says his system uses as many as 14 cameras positioned around an actor to get a 3D picture of their movement in real time.
But Natal comes equipped with just two cameras. So how can it work with such a limited input?
In-depth view
A year ago, Israeli firm 3DV Systems unveiled a system that uses an infrared depth-sensing camera. The ZCam is said to be able to pinpoint the depth of an object to within 1 to 2 centimetres, and capture the information at a rate of 60 frames per second for very smooth motion. A second, full colour camera in the device records textures and colours.
In March of this year, the company sold its assets to a third party, reportedly Microsoft. Many people have concluded that Natal's motion capture system owes a lot to the ZCam. When New Scientist asked 3DV System's CEO Zvika Klier about the possibility this week, he had no comment.
Tschesnok says the results from ZCam-like systems fall short of what he'd call true markerless motion capture – Natal can't see what's going on behind a gamer's back – but it is still an improvement over other games consoles' interfaces and will work well for Natal's needs. "In a way, it's the same effect as having a Wiimote in each hand, one on each foot and one on your head," he says.
But more is yet to come: Tschesnok predicts that the games consoles due out early in the next decade will use his multi-camera system, built into the widely used "surround sound" speaker systems to provide users with a truly immersive experience.
Voice in the crowd
People are also understandably excited about Natal appearing to recognise and respond to voices and emotions. However, while speech recognition software may be increasingly common, it still suffers from the "cocktail party" effect; where background noise and multiple voices can cause errors.
Finding a way round this issue is vital for gaming. Microsoft is targeting Natal at social and casual gamers who are likely to play in the company of others, yet few, if any, voice recognition systems have been created capable of following a single voice in a noisy room.
IBM has shown that a system that can read lips provides one way around that. That could help Natal, but would need to be tuned separately to different languages.
Research that enabled one of Honda's Asimo humanoid's to understand three voices all speaking at once even suggests that multiplayer voice control could be possible one day.
Face reading
Speech software could also feasibly be used to recognise emotions from stresses and emphasis in a player's voice. For example, software called Emotive Alert is able to classify voices into one of eight basic emotional states.
We are told, though, that Natal will use facial recognition, so it may read emotions that way too. Earlier this year, artist Tina Gonsalves teamed up with neuroscientist Chris Frith at University College London, UK, to develop an art installation that responds to the emotion of visitors, using an algorithm Frith created to read faces.
Frith told New Scientist that getting such systems to perform reliably is difficult, but not impossible. However, it is unlikely that Natal will detect more than a handful of basic emotions.
Emotional connection
Even with that limitation, a console able to sense emotions and respond to them has the potential to make gaming a much richer experience.
In 2005, an "emotionally aware" virtual fitness trainer, name Laura, was tested on groups of volunteers. Her friendly gestures and sympathetic body language were found to genuinely foster a better connection with users. This positive influence significantly increased the participants' exercise levels compared to a control group interacting with a version of Laura that didn't recognise emotions.
content ads
Tuesday, August 11, 2009
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment