Smart Guns Stall
Despite a new law, technology that lets only a gun's owner fire it is suffering from anemic research funds and industry intransigence
3 September 2003—New Jersey lawmakers can be labeled optimists, catalysts or fools when it comes to so-called ”child-proof” gun technology. Bolstered by state-funded biometrics research done at the New Jersey Institute of Technology (NJIT, Newark, N.J.), lawmakers got a popular but controversial, first-of-its-kind bill signed into law in December. The bill will make user-recognition technology for handguns mandatory once it has been developed. The state’s attorney general has been given the authority—though not yet the criteria—to decide when the technology should make the leap from the lab to the shelf. But now, eight months after it went into effect, little progress has been made toward commercializing smart guns, as research suffers from a lack of funds and an unethusiastic gun industry.
Work on user-specific or personalized gun technology began with mechanism that require users to enter a PIN code or wear a transponder (a ring or wristaband) to unlock the trigger. While some researchers are still trying to refine those techniques—especially for law enforcement use—NJIT and others are looking to design a smart gun for the common consumer with bump-in-the-night defense concerns. To make a smart gun simple to use, researchers have turned to biometrics, the science and technology of measuring and statistically analyzing biological data. A few companies have looked at fingerprint scanners for handguns, so far with limited success. Though researchers at NJIT say they are still open to all ideas, they have bypassed more established techniques in favor of an experimental biometric approach they call dynamic handgrip recognition. By measuring the pressure with which a person grips a gun in the milliseconds before firing, researchers hope to find a recognizable and unique signature.
A few big-name gun manufacturers have pledged to work with NJIT on the project, but even these say they have scant resources to devote to research and development. They point out that their legal fees in recent years have risen as cities have launched lawsuits claiming, among other things, that manufacturers have been negligent in not designing safer guns. Even with the full cooperation of handgun makers—certainly not a given—researchers at NJIT will need to push the limits of technology and overcome a substantial funding shortage to develop a working prototype.
Getting a grip on gun violence
Since 1999, NJIT researchers have been exploring their dynamic handgrip theory. They took photos of more than 160 people gripping and firing a mock Glock handgun. Based on the photos, researchers decided where to place shirt-button-sized piezoelectric pressure sensors on the grip of an actual gun. Designers expect the piezoelectric sensors, which generate an electrical charge when mechanically deformed, to also help power a signal processor and the memory needed to make the gun programmable.
Dynamic handgrip recognition has several advantages, according to NJIT researchers. For example, they say that, unlike fingerprint scanners, their method should work in harsh environments and in situations when the user is wearing gloves.
But other biometrics experts expressed doubt. Lisa A. Osadciw, a Syracuse University (New York) assistant professor of electrical engineering, called handgrip recognition ”one of the kind of ’out there’ technologies esoteric.” Anil K. Jain, a Distinguished Professor at Michigan State University (East Lansing), said ”I don’t think there’s any evidence that suggests [grip pressure] is unique to an individual To me it doesn’t sound credible.”
Neither Jain nor Osadciw have researched biometric applications specific to handguns. But to improve reliability rates of biometrics in general, both have focused on a multi-modal approach, combining fingerprint scanners with other techniques such as face recognition.
NJIT researchers point to preliminary testing done at the Picatinny Arsenal in New Jersey last year as proof of the promise behind handgrip recognition. Making use of a simulator designed by the military to mimic the force feedback of live-fire conditions, six U.S. Army marksmen squeezed a gun grip embedded with pressure sensors. Each marksman fired the gun for five trials of 20 shots each. Two marksmen even wore gloves for two of their trials. Then using a statistical method known as bootstrapping, which involves a repeated, random resampling of data, NJIT researchers found that by examining the readings from the pressure sensors, they could correctly identify which of the six marksmen was firing the weapon about 95 percent of the time.
Since that trial, researchers have refined their patented design. They’ve increased the density of the pressure sensors in the gun grips and integrated into the grips a printed-circuit board that does some signal amplification and filtering, though the digital processor is still a separate unit. Researchers plan to go back to the arsenal in September to do further testing, and hope to place their grips on a live-firing weapon in October.
They have a long way to go to match the best, most expensive biometrics equipment—and even those cutting-edge technologies still have an error rate of about 1 in 1000. And while most shooters know that a misfire happens not infrequently, manufacturers and shooters will be hesitant to accept any smart gun that isn’t smart enough to let its owner fire it. This false rejection rate must be no greater than the mechanical failure rate attributed to handguns, about 1 in 20 000. ”It has to be reliable,” says Smith and Wesson (Springfield, Mass.) marketing director Ken W. Jorgensen, adding that there will be ”no compromise, working nine-tenths of the time.”
But by tweaking the algorithm that takes the sensor data and determines that the hand gripping the gun belongs to an authorized user, NJIT engineers can make tradeoffs between false acceptance and false rejection rates. Their current target is to screen out all but one in a thousand unauthorized users. They say that even if the technology can prevent only 99 out of a hundred unauthorized users from firing the gun, it could be considered a success. (In a so-far untested theory, children with smaller hands than adults might be screened out more easily using dynamic grip recognition.) Whether that will be good enough to sell consumers on the idea is an open question.
Money and politics
On 29 August, NJIT scored a commitment to jointly commercialize a gun under development by Metal Storm Ltd. (Brisbane, Australia). Metal Storm is a research and development firm rather than a handgun maker, but it has made a prototype of an all-electronic handgun. Rather than firing bullets using a mechanical mechanism, the triggering is controlled by circuitry in the grip. ”[The circuitry] allows a range of ’personalizing technology’ to be incorporated in Metal Storm’s original design of the weapon. It’s not an ’add on,’ which has been the difficulty in the past,” says Donald H. Sebastian, vice president for R&D at NJIT.
It seems unlikely, however, that in a few years New Jerseyites will be buying all their handguns from a small Australian company. But while an R&D firm like Metal Storm excitedly puts out press releases to announce its partnership with NJIT, the university has received mixed signals in its courting of traditional gun makers.
To be sure, manufacturers recognize that some people, afraid to own a gun now for fear of accidents, might purchase one if it were safer, boosting sales. However, making a gun more complex with an ”add-on” biometric trigger lock—even if technically feasible—would no doubt increase costs, possibly hurting profits. The goal of the NJIT researchers is to install the trigger lock without adding more than 10 percent to the cost of a gun. But the feasibility of this figure has already been contested by some gun manufacturers. While the researchers said users could reasonably expect to pay about $50 more for smart guns, Beretta USA Corp. (Accokeek, Md.) has tabbed the extra cost at about $300.
Beretta has been outspoken in its opposition to smart gun technology, lobbying against a smart gun mandate that was considered in Maryland in 2000. Then-governor Parris N. Glendening reportedly proposed $3 million in funding to in-state gun manufacturers to develop the technology. But Beretta remained unswayed, saying it had done research on its own and concluded that the technology was not feasible and, in any case, it was too expensive and unnecessary when mechanical gun locks already exist.
In contrast, five companies—including big-name gun makers Smith and Wesson, Taurus International Manufacturing, and Colt’s Manufacturing Company—signed on for a $2.7 million grant proposal sent to the National Institute of Justice back in the summer of 2001. Despite agreeing to work with NJIT as part of the grant, Smith and Wesson remains cautious about predicting success for any smart gun technology. ”We have no guarantee of the deliverability of such a product,” says Smith and Wesson’s Jorgensen flatly. He says that the company has spent millions of dollars on research into smart guns.
Colt has also done work on its own. It tried unsuccessfully in 1999 to launch a start-up it dubbed i-Colt to work solely on smart-gun technology (focusing mainly on token-based technology like a radio-emitting wristband). The company blames, at least in part, the cost of liability litigation for the failure of its smart guns project. Carlton Chen, general counsel and secretary for Colt, testified before the U.S. Congress in April that ”all our money is being diverted to defend ourselves in these lawsuits and we have had to slow down in our smart gun technology development.”
This claim, a familiar one to NJIT researchers, was weakened this year when an internal document became public during a lawsuit involving Colt. A June 1999 memo suggested that Colt decided not to make public their true progress in developing a smart gun in order to encourage more government funding for the technology. According to the memo, at the time Colt thought a law enforcement model was two to three years away, while a consumer model was thought to be possible two to three years after that.
The excuse may be all but eliminated soon, as the gun industry is trying to push a bill through Congress that would shield them from such suits.
When passing the New Jersey bill, state lawmakers talked about how their mandate might speed the development of a smart gun, but NJIT has yet to receive any additional funding. On the heels of the New Jersey law, NJIT researchers said they needed $5 million and three years to build a prototype, with institute researchers developing the user-recognition technology and gun manufacturers working out the mechanism that would prevent the gun from firing.
Without additional funding, NJIT researchers say they could stretch the $1 million in state research funds they have already received to continue their work until 2004. Despite their current funding woes, though, the researchers believe that as long as they have the continued interest of legislators, smart gun technology can’t miss.