Dr Arkady Zaslavsky, associate professor at the School of Computer Science and Software Engineering at Monash University supervised student Daniel Hagglund in undertaking the project, which studied the feasibility of "tracking and to some degree predicting the movements, direction, pace of an object, for example, a human that makes some sound...says a word or a sentence."
"The main aspect [of the project was] to address some aspects of pervasive computing, namely, use the ability of mobile software agents to track objects of interest, and in this particular case, using sounds those objects make," Zaslavsky told ZDNet Australia .
Pervasive Computing, which IBM defines as any computing device that is not a personal computer, is predicted to take off over the next few years as everything from cars to fridges to desk lamps have computer chips installed and be connected to the Internet.
"We attempted to demonstrate the feasibility of tracking sources of sound with the use of mobile software agents if those agents have access to available resources, spare computers with active microphones," said Zaslavsky. "One possible application, for instance, is to let the user's e-mail window follow or run ahead using computed, expected, typical, [or] whatever route the user takes."
In this scenario, a home-based worker who left the office to go to the kitchen to get a snack could have the document they were working on be displayed automatically on a screen in the kitchen. Of course the same technology could be applied to have a real-time computer game follow you about the house.
The demonstration and implementation of the program were done on two IBM computers in different rooms, using Voice Micro-edition.
"Tracking is done with the use of mobile software agents, that is, pieces of software that can migrate between computers on their own volition or according to some itinerary [and] have access to sound recording devices such as microphones," said Zaslavsky. "Once those agents detect a sound and match it with the recorded pattern of the trackable object and build the object's trajectory, they clone themselves and roam along the expected route and wait for the same sound and so on."
"The agents locate the user by measuring volume of the recognized sound and the agent (hence computer where it resides) which reports the loudest sound is the approximate location of the trackable object with some accuracy/radius," explained Zaslavsky. "We didn't do triangulation though had plans for it. Since we have a floor map with computer locations we can approximate the trajectory of the object."
