You can now operate your computer from a distance of up to 4 metres, and that too, with just a finger, thanks to the innovative thinking of three smart students at MIT, Pune.
Stand in front of your laptop, flick your wrist to your left and your search screen suddenly returns to the previous page. Or, make the gesture of a crossing-out sign to close an application. Seems like a scene right out of the Harry Potter series, right? Nah, trust me,it's real. Three students, Swapnil Vispute, Aniket Nagle and Madhavi Paropkari have made this magic possible, under the supervision of Professor S.S. Paygude at the Maharashtra Institute of Technology (MIT) in Pune. Christened Touchless Interface Environment or TIE, the interface they have developed can comprehend a single-hand gesture in real-time.
"The latest technology in the field of human-computer interaction includes intelligent interfaces like touchscreens. Our dream was to go one step ahead of what is available currently "to build an interface in which you can control your computer system without touching any device, but just by hand gestures in mid-air," says Swapnil Vispute.
"Since this was the idea of students, and has never been implemented before, it was a gamble on the part of MIT and our sponsor Celunite Inc. But their constant support and guidance has brought us this success. Without them, touchless interaction would still have been a dream!" he adds.
How does it work?
The system just requires a desktop (or laptop) and a webcam. Even a webcam with a low resolution works fine. Everything else is managed by a software unit called TIE (touchless interface environment). What's more, the whole unit is built on an open-source platform. Since open source is the next big thing, we decided to implement TIE on Fedora Core 6, kernel 22.214.171.124, using the gcc compiler. We have used the C language for coding. The project code and all the software components are FREE, and the only cost incurred is that of the webcam," puts in Madhavi Paropkari.
TIE has two parts: the image processor and the driver. The image processor makes use of a core algorithm called CAMshift (Continuously Adaptive MeanShift), which applies various image-processing functions to determine the position of the hand and the gesture made by the user. This information is then passed to the driver, which synchronises the position of the hand with training information stored in computer memory, links them together, and performs the action. For example, when a user motions a crossing-out gesture, the webcam captures the image and sends it to the image processor. Here it is cleaned of noise, and a histogram equalisation is performed. The classification algorithm then links the image to a pre-defined set of actions and classifies it. Once that is done, the drivers object-tracking mechanism tracks the position of the hand and moves the cursor accordingly, and the application closes down. Thus all functions like a single click, double click, cursor movement, scroll, and drag-and-drop are done by just making hand gestures in front of the webcam, without touching any device.
"When we initially developed the product, it could only perform limited functions. The keyboard was still required. However, now with improvisations we have eliminated the need for a keyboard too. The TIE system thus completely replaces the mouse, joystick and trackball, and can be extended to perform keyboard functions," says Vispute.
Conceptually, TIE seems quite unique--a touchless interactive interface. At the brick-and-mortar level, it still has a few hiccups, like consumption of excessive power and memory. Other challenges include real-time image processing and simultaneous interaction with the host operating system, be it Linux or Windows. "They are working on it, and very soon these concerns will also be solved," says Professor Paygude.
Seeking commercial success
Given its potential, it's hardly surprising that a number of industries are showing interest in TIE. "The innovation has immense capabilities, and will serve as a boon for a number of industries. From gaming to medicine, it will find acceptance everywhere," says Deepak Ayer, CEO, Celunite Inc.
"During operations, doctors need to use computers to monitor vital parameters, refer to old patient data, or to an old case that might be useful for the present operation. In such cases, they cannot afford to waste time washing hands and using the keyboard and mouse interface to access such critical data. Instead, a touchless screen would provide the doctor all the information he wants during his operation," puts in Aniket Nagle.
TIE has other applications too. It can be used to control household electronic devices like lights, air conditioners, televisions, etc. "All that has to be done is to interface the TIE system with these devices. However, the challenges for such a system will be that of identifying the current user and training the system for multiple hand gesture recognition," puts in Vispute.Even as this goes to print, Vispute informed us that the group is also trying to add a DSP co-processor to the system, in order to work towards integrating voice recognition into the TIE system. If all goes well, we can bid goodbye to our keyboards, mice and remotes forever.
While the TIE system may seem similar to prototypes already developed internationally, it is different in more ways than one:
- The TIE system does not use the EFS (Electric Field Sensing) technique. In EFS, an electric field is suspended in front of the screen with the help of rod-shaped transmitters and receivers. When an object, like your hand, comes near the screen, it interrupts the electric lines of force. This causes a change in the electromotive force received at the receiver. The horizontal receiver gives the 'co-ordinate' and the vertical receiver the 'co-ordinate', thus locating the point where your hand entered the electric field.
- The TIE system is very cheap, since the only cost incurred is on the webcam. EFS-supported screens are very expensive.
- The TIE system allows the user to operate it from a distance of up to 4 metres; this distance can be increased by using a higher-resolution webcam.