Air Force wants AI for cislunar space vehicles

The Air Force is surveying industry for ready-made artificial intelligence and machine learning software geared toward autonomous space vehicles operating outside of the near-Earth domain, a new document says. 

The service is seeking out “flight-ready” concepts to help cislunar space vehicles run through autonomous maneuvers like “station-keeping, trajectory optimization, and hazard avoidance” of objects moving outside of planetary orbit, according to a request for information published Monday. 

“The Government has an over-arching policy objective to develop a sustained architecture beyond earth orbit,” the RFI states. The replies, due Dec. 4, should drive future requirements, it says. 

Cislunar systems deal with a much larger domain volume than that of near-Earth systems, which can cause communications delays and issues for command and control systems on the ground, according to the RFI. 

“Communication latencies that result from the increase in volume may mean that traditional, ground-based, command and control and mission planning systems are no longer scalable on operationally relevant timelines,” the RFI states. 

The Air Force is therefore asking for industry expertise in the form of white papers that should outline concepts for the cislunar domain around five topics: autonomous mission planning, predictive analytics, trajectory planning, computer vision and other potential technical areas, according to the RFI.  

The service wants fully developed autonomy software that would require no development and little training inside of 90 days of an “Authority to Proceed” order and to be fully qualified within a year and a half, per the RFI. 

The RFI follows comments from Tom Ainsworth, principal deputy assistant secretary for space acquisition and integration, who said in March the Space Force would start standing up leadership positions to acquire technologies to target the cislunar mission area. Ainsworth was performing the duties of Air Force assistant secretary for space acquisition and integration at the time.  

The RFI also asks for industry’s take on the potential hardware, compute and data requirements for their AI/ML algorithms, including “any anticipated data rights considerations.”

The information gleaned from industry may yield Air Force investment in future studies around the concept, although the RFI stresses that the service wants to see industry’s ability to deliver readily available software fast. 

The government “is not interested in investing in a cislunar SI research and development program,” it says, using the Trump administration’s preferred term for AI.