Aptima has won a contract from the US Air Force Research Laboratory (AFRL) to develop an automated librarian that uses ‘imperfect’ artificial intelligence (AI) for pilot training.

Valued at more than $5.2m, the four-year contract was awarded by the AFRL’s Airman Systems Directorate, Warfighter Interactions and Readiness Division.

It also includes financing for the Gaming Research Integration and Learning Laboratory (GRILL).

GRILL is said to provide a typical environment for testing new technologies and their capabilities.

The latest contract will see Aptima manage seven developers of AI agents, who will establish artificial enemy pilots capable of conducting aerial missions against human pilot trainees.

The developers include CHI Systems, Eduworks, Soar Technology, Stottler Henke Associates, TiER1 Performance Solutions, Charles River Analytics and Discovery Machines.

How well do you really know your competitors?

Access the most comprehensive Company Profiles on the market, powered by GlobalData. Save hours of research. Gain competitive edge.

Company Profile – free sample

Thank you!

Your download email will arrive shortly

Not ready to buy yet? Download a free sample

We are confident about the unique quality of our Company Profiles. However, we want you to make the most beneficial decision for your business, so we offer a free sample that you can download by submitting the below form

By GlobalData

According to Aptima, the programme’s previous phase involved the development of several AI pilot agents with each of them showing a substantial variation in performance across different scenarios.

Based on this difference, the company seeks to develop an automated librarian that can personalise learning and training for pilots.

The new concept categorises the AI pilots to smartly matches them to pilot trainees.

Aptima chief scientist Jared Freeman, overseeing the AI programme, said: “For training to be effective, it should be neither too difficult, nor too easy, but in the sweet spot for the trainee to learn and develop their competencies.

“Much like training in sports, where you want to practice against an opponent that stretches and develops your game, an AI agent chosen well by the librarian will be just better enough than the trainee for them to learn how to win, pushing them just to the bounds of their learning envelope.

“It is usually ineffective to have a dominant adversary that repeatedly overwhelms a novice pilot or a weak adversary that doesn’t challenge the trainee to improve.”

In addition to developing the automated librarian, Aptima will create a Cloud-based testbed that scales to allow AI agents to simultaneously generate flight data.

In April 2018, Aptima and Lockheed Martin Aeronautics (LM-Aero) partnered to develop a new system that can monitor complex spaces in the USAF and other defence depots.