Pentagon Shops For AI Computing System For MQ-9 Smart Sensor

Credit: U.S. Air Force

A U.S. Defense Department agency is shopping for a computing system powerful and efficient enough to process and interpret hundreds of teraflops of imagery data from systems on board manned and unmanned aircraft.

A “sources sought” notice released on Aug. 17 by the Joint Artificial Intelligence Center (JAIC) starts market research for an edge-based, high-performance computer that can be installed initially on a General Atomics Aeronautical Systems Inc. MQ-9 Reaper. 

The MQ-9 is being equipped with the JAIC’s Smart Sensor technology, which requires an extremely powerful computer to sort through more than 500 teraflops—defined as a million million operations per second—contained in imagery data to identify targets and other points of interest. 

The JAIC also plans to install the Open Mission Systems-compatible computing system on other UAS and manned aircraft within months or years, the sources-sought notice said. 

The computing system could come from an existing government system or a commercial product, according to the JAIC notice. But the challenge will be finding a system small enough to package within the size, weight and power limits of an MQ-9.

“The primary competitive factors for consideration are, in order: compute density, scalability, composability, upgradability, total cost of ownership, bandwidth, latency, and security,” according to the notice. 

The system also must be designed to survive in an MQ-9’s nominal mission conditions at 20,000 ft. with minus 8C ambient temperature.

Steve Trimble

Steve covers military aviation, missiles and space for the Aviation Week Network, based in Washington DC.