AI-Powered Computer Could Tailor Training

iStock illustration

ORLANDO, Fla. — A new data engine that its creators say can be paired with any simulator made its debut at the National Training and Simulation Association’s annual confab in November.

The extended reality analytics engine, or XRAE — displayed by developer Booz Allen Hamilton at the Interservice/Industry Training, Simulation and Education Conference in Orlando, Florida — uses algorithms to compare information collected from a simulator to data standards set by any military service in real time.

“The services are all trying to really increase their readiness with less resources,” said Elizabeth Robinson, principal and director at Booz Allen Hamilton.

Every time a service member puts on a headset or grabs a controller for a simulation event, those tools generate a huge volume of data, she explained.

However, much of the data isn’t being utilized. For “most of it — nothing’s being done,” Robinson said. “It’s sitting there, or they’re pulling it out and applying a whole lot of man-hours to look at it.”

Because the engine processes data in real time, the performance evaluation is ready immediately after a training session. This can cut down the time and resources needed to determine the best learning strategy for an individual.

Robinson pointed to the speed at which U.S. adversaries have adopted advanced technologies. For example, data officials at the conference noted China’s advanced artificial intelligence capabilities.

“We used to have decades to prepare for threats,” she said. “The world’s a lot faster place than it used to be.”

Moreover, the machine learning and other technologies that power the engine have only recently become accessible to the government.

“The technology is just now being able to be leveraged ubiquitously, because some of the services have tech adoption challenges,” said Richard Ayers, senior lead human performance engineer at Booz Allen Hamilton.

Accessing training data can benefit senior officials, who can explore where training needs to improve, he said.

During a demonstration, a sensor tracked movement during a flight simulation. The engine calculated how often the user’s altitude and pitch matched an ideal flight pattern and displayed the information on a dashboard.

Determining training performance for different objectives can help leaders distribute resources, Ayers said.

“I’m not having to ask for more money,” he said. “I’m just being really, really smart with evidence-based decisions on the resources that I do have.”

Robinson noted the product is being used in two different locations by customers, but said the company is not permitted to provide more detailed information.

Topics: Training and Simulation