Among the many tasks assigned to Transportation Security Administration (TSA) Transportation Security Officers (TSOs), they must screen every bag boarding commercial aircraft within the United States.
The contents of bags are displayed on a screen as the scanned items pass through an X-ray machine. TSOs must identify threats, while minimizing unnecessary manual secondary bag searches that slow checkpoint throughput when they mistakenly flag a benign item. Interpreting X-ray images and understanding what threat and non-threat items look like in innumerable orientations is a daunting visual task.
Visual search of X-ray images is a repetitive task for the approximately 50,000 screeners employed by TSA, with an often low probability of encountering a threat. TSOs are trained to use perceptual cues such as color, orientation and spatial location of individual items to identify potential threats and differentiate them from non-threat items in the X-ray images of scanned bags.
The huge volume of items scanned every day must be cleared as quickly and efficiently as possible to facilitate air travel. However, missing a threat could be catastrophic. Lives depend on TSOaccuracy. TSOs battle competing demands for safety and speed daily.
S&T says that the Department of Homeland Security Science and Technology Directorate’s Office for Public Safety Research (OPS-R) developed a training system that not only makes TSOs more efficient, but also maintains their accuracy.
Existing training software is limited and uses only exposure training to elicit improvements in threat detection. Current training uses example after example until a TSO becomes more proficient. This training method is not adaptive to an individual’s needs, does not leverage the latest training methods or technology, and does not identify the root causes of a TSO’s deficiencies.
With these needs in mind, OPS-R sought to develop TSO training methods and tools that not only leveraged innovative emerging technology, but would also be relevant, challenging, intuitive and engaging. Enter ScreenADAPT, an advanced X-ray image analysis training system that examines TSO performance based on the latest in visual search research and uses eye-tracking technology to examine visual search performance.
ScreenADAPT has two main advantages over traditional training methods.
First, the program has an eye-tracking capability. Right now, trainers and trainees do not have any objective measure of where they are looking on the screen, for how long and in what pattern. This can all be determined using ScreenADAPT. This critical information allows trainers and trainees to analyze why an error was made. It can now be determined if a threat was missed because a trainee did not look at that area of an image, or if they looked at that area and did not recognize the item was a threat.
Second, ScreenADAPT provides diagnostic metrics on TSOperformance and automatically adapts to the trainee’s needs. ScreenADAPT dynamically addresses the trainee’s needs by varying the type of training, type of threats, level of bag clutter and difficulty.
“You make this type of error, you get this type of corresponding training. If they are missing guns, they will see more guns; if they are missing IEDs they will see more IEDs,” said Darren Wilson, OPS-R’s ScreenADAPT program manager. “If a TSO makes a scanning error, they will get more exposure training. But if they make a recognition error, they will receive a different type of training to combat that, called discrimination training. The different types of training address the corresponding root causes of the errors and assist in building each TSO’s mental threat image library.”
The initial effectiveness evaluation indicated that using ScreenADAPT in initial training not only resulted in TSOs identifying threats faster, but also clearing bags faster. They were able to make faster decisions with more confidence.
Customization at the airport level is also a major advantage of ScreenADAPT. The items passengers pack in their bags in Portland, Oregon, at any given time of year can be very different than the items people pack in Orlando, Florida. ScreenADAPT allows individual airports to upload their own images that reflect the threat environment and items most often seen in that locality, as well as be responsive to emerging threats.
ScreenADAPT can be configured with a single or dual screens, and includes a small camera located just below each monitor to unobtrusively track eye movements.
The ScreenADAPT software calculates all of a TSO’s hits, misses and false alarms, and it can compare their performance metrics to those of their peers. The feedback ScreenADAPT produces helps improve all scanners’ performances.
“The whole premise is to maximize visual search performance,” said Wilson. “What this enables you to do is to train the core visual search skills necessary to efficiently and effectively conduct X-ray image analysis. Even if there are advancements in technology, it’s always going to come back to these basic visual search skills.”
TSA has recently deployed fifty ScreenADAPT systems for an even larger training effectiveness evaluation at airports in New York City, Pittsburgh, Portland (Oregon), Houston, Las Vegas and Raleigh. Some of these locations volunteered to be research airports, while others were chosen to vary the size of airports and types of passengers in data collection.
Initial data indicates that ScreenADAPT results in a 45 percent improvement in efficiency, with no loss in threat detection effectiveness.
For more on U.S. Airport Security, Homeland Security News Wire