By Calvin Biesecker
UPPER MARLBORO, Md.–After a year or so of initial development and integration work, the Department of Homeland Security’s (DHS) Science and Technology (S&T) branch this week began a demonstration of sensor and data fusion technologies that one day will automatically detect if individuals are contemplating hostile actions based on biological changes that may or may not be visible to the naked eye.
The Future Attribute Screening Technology (FAST) program is measuring changes in a person’s biological attributes, such as their body heat, skin presence, breathing patterns, heart rate and micro-facial expressions, to determine malintent, Robert Burns, the FAST program manager within S&T’s Human Factors Division, said at briefing about the experiments. Measuring physiological and behavioral attributes avoids random profiling, he said.
The data is being collected via sensors from stand-off ranges–several feet or more in the current experiment–and then analyzed by sophisticated algorithms that are still being developed to ultimately fuse the various sensor outputs to determine if a subject may mean to do harm.
FAST is a decision aid that would help a security guard or officer make a determination, Burns said. However, humans get tired or their attention wanes, whereas the technology being integrated into FAST doesn’t, Sharla Rausch, director of the Human Factors Division, said.
DHS initiated the FAST program about 15 months ago as a high-risk, potentially high-payoff attempt to see if technology can be used and or developed that can aid security officers–typically those who would man some type of checkpoint or venue entrance–in determining whether a given person should be pulled aside for additional screening. That additional screening might consist of a security officer asking questions and more sensors being used to help determine if a person does in fact pose some sort of threat.
As one of S&T’s Homeland Innovative Prototypical Solutions efforts, FAST program officials are trying to deliver prototype demonstrations that result in “game changing” technologies over a two to five-year period.
Ultimately DHS S&T hopes that the technologies developed and integrated under FAST contribute to higher throughputs of people into a venue or through a checkpoint and more valid referrals for secondary inspection. While current testing is being done in the mobile lab, eventually such a lab could be used for special security events but in other applications the sensors would be part of the fixed infrastructure.
For this week’s testing, which is being done at an equestrian center in a semi-rural area of Maryland outside of Washington, D.C., 140 test subjects are being shuttled one-at- a-time through the FAST mobile lab. The subjects only know they are part of a security screening demonstration, although some are being given “items” to smuggle so that as they are being queried by security officers at primary and secondary inspection points inside the lab, their bodies physiological reactions are being monitored. Program researchers are then using that data to create baselines against which they can measure the performance of the sensors and data fusion algorithms.
This fall S&T plans to do additional FAST tests in the lab with 200 test subjects–which is an amount that provides a statistically relevant sample–to further enhance their data algorithms and sensor systems. Then again next spring, the test will be repeated with 200 more subjects.
In the 2009, 2010 time frame is when more robust data fusion efforts will take place. Right now for the demonstration it takes about a minute for the sensor outputs to be analyzed, which currently requires a person in the loop. Eventually the goal is real-time, fully automated analysis.
For the current tests the FAST mobile lab is equipped with a thermal imaging camera provided by FLIR Corp. [FLIR] that is combined with software to analyze the changes in a person’s body heat as he or she stands still while a primary security officer asks specific questions. In addition, a light detection and ranging radar system, balled the BioLIDAR that was developed by Digital Signal Corp., is used to observe surface changes on a person’s neck or face so that respiratory and cardiovascular patterns can be monitored and measured.
As the program moves along, S&T will utilize additional sensors and technologies.
The mobile lab, and a related command center, was integrated by Battelle Memorial Institute under a one-year, $1.7 million contract. Battelle also acquired the test subjects. Draper Laboratory is working under a one-year, $2.6 million contract to do the data fusion for all of the sensors. Both contracts are currently being negotiated for extensions.
Future experiments under FAST will also look at using a camera that is integrated with algorithms to see how well technology can be used to observe micro-facial expressions– which are over and done with in an instant on a person’s face–to create another input into the data fusion algorithms so that the ultimate system being developed can more accurately predict intent.
As for whether technology can be used to automatically interpret micro-facial expressions for behavioral intent, “it’s possible,” Paul Ekman, the leading authority on micro- facial expressions, told Defense Daily yesterday. But it will take investing money into the research that so far hasn’t been done, he said. Ekman has provided DHS advice with on the FAST project.
Other attributes that will be tested in the future include gestures and other body movements, Burns said.
S&T’s Human Factors Division has also awarded some small contracts to possibly bring other technologies into the FAST testing. One is a chemical sensor being developed by researchers at the Univ. of Pennsylvania that can sniff out odors on a person. Those odors change depending on a person’s emotional state, a Penn researcher told Defense Daily.
Under another contract, Veridical Research and Design Corp. is investigating how changes in a person’s pupils can be used to show deception. When someone is under stress or is thinking hard, their pupils dilate. Veridical is trying to model if someone’s pupils are dilating because they are thinking hard or because they are nervous, a company official said.
The foundation behind FAST is a relatively new theory called Malintent. There are various cues that underpin malintent, such as cardiovascular, respiratory, thermal, body movements, and more, Dr. Jennifer Martin, a consultant with MRAC who helped develop the theory. FAST combines the cues to determine of malintent exists, she said.