Using a listening device to identify direction of gunfire
Using A Listening Device To Identify Direction Of GunfireIn an unusual application of neuroscience research, police agencies around thecountry may soon be able to equip street corners with microphones and videocameras to fight gun-related crime.The system uses the equipment and a computer to recognize gunshots, pinpointwhere they came from and transmit the coordinates to a command center. It relieson software that mimics the way the human brain receives, processes and analyzessound. The system has drawn the attention of several law enforcement agencies,including police departments in Chicago, Oklahoma City and Phoenix and theLos Angeles County Sheriffs Department."When you put in automated gunshot recognition in a highly visible format likethis, the residents no longer have to fear reprisal and the police no longer have todepend on the residents for accurate information.The system was able to distinguish gunshots from voices, car traffic andconstruction. "You can find that brains can do it, but you cant find physicalsystems that can do it. "Its very hard to corrupt the signal in such a way that wedont know whats going on."The software is able to recognize sound by dividing it into small sections andmatching them to known audio-wave patterns through a technique known aswavelet analysis. The system also examines the incoming sound as a whole,making sure that its components are heard in the proper order, even if one isobscured by other noises, such as a loud truck.Over time, the system can learn to recognize certain sounds. "The gunshot has aparticular unique signature, where you have the sound of the explosion when thefiring pin hits the bullet and the sound of the bullet as it expels from the gun."Youve got a little blip, and then a drop, and then a big blip, and with the trainingweve done now, weve picked it up so that both blips are part of what it haslearned."
The system uses four microphones, contained along with the computer in abulletproof box two feet by two feet by three feet. The system is able to determinequickly where a gun was fired, using the difference in time that the sound took toreach each microphone. The computer sends a signal to the video camera, whichzooms in on the location. The system then transmits information and the videodirectly to law enforcement headquarters. The devices, which cost up to **** each,can cover an area with a radius of about 200 yards.The Los Angeles County Sheriffs Departments person who is in charge ofevaluating technology for the agency, said that the system is likely to providegreater "situational awareness." In about six months, the sheriffs department willtest the devices, installing as many as 20 around Los Angeles County.The information and video images can be immediately reviewed by officers atheadquarters who may be able to determine the nature of the incident, he said."Because of that we gain a tremendous tactical advantage before we arrive on thescene," he added.Beyond standard urban law enforcement, the American military has taken notice ofthis are of research. In fact, the Office of Naval Research in Arlington, Va., hasbeen financing this research for the last 10 years.In the summer of 2002, the inventor had been working on applying his research tovoice recognition software, and discovered that it worked even in very noisyenvironments.The Office of Naval Research and other agencies approached the inventorindependently to discuss possible commercial applications. He said that themilitary had ideas for using his technology to monitor "security-breaching noises,"such as the sound of a chain-link fence being cut, or to recognize human voices inunauthorized areas."It has a lot of applications," said the program officer at Office of Naval Research."Any perimeter that you want to secure and you dont want to expend a lot of assets
in guarding - you want to do it on a 24-hour basis - these systems are all good forthat."The program officer at Office of Naval Research envisions the day when thesystem can be incorporated in a hand-held device that soldiers could use in thebattlefield.A professor of neurosurgery at the University of Pittsburgh and a member of acommittee that reviews this research, said that his work on voice recognition was"a major step forward in terms of what people have done." The gunshot-recognition system is based on the same technology, and while it is possible that itwould not work as well, "I dont know why it wouldnt," he said.The inventor said that the next version of the system might also include the abilityto recognize different sounds at once. For example, the ability to recognize ahuman voice, human footsteps, gunshots and the climbing of a chain-link fence,which in combination would paint the picture of an armed break-in. He also wantsto upgrade the systems visual abilities so that it could track the shooter.To do that, the system would have to recognize human shapes. "There are multiplethings in an urban setting that can be moving," he said, "but we want to have formidentification.