The Problem This Solution Solves
The Air Force relies on historical test and simulation data to train models that can accurately predict whether an aircraft can operate safely when loaded with new sensors, weapons, electronic warfare equipment, and other devices. This data is held in millions of records locked in disparate silos, is manually processed, and is rarely used to its full potential, all of which slows down the certification process while increasing workforce constraints.
In 2018, DIU partnered with Tamr, a leading machine learning and data mastering company, to develop analog testing models that speed up the configuration assessment process and reduce engineers’ manual workloads. Tamr accomplished this task by leveraging databases of previous configuration information to build models that could identify promising new configurations that met mission parameters. In doing so, Tamr was able to reduce both the time required for and the cost of test flights and simulations for new aircraft configurations, ultimately increasing mission capable rates to 80% across multiple USAF air platforms.. The initial prototype focused on the F-16 Fighting Falcon and has since expanded to other aircraft platforms as part of the production contract.