/

Army-Led AI Study Identifies Collaborative Multiagent System Development Approach

2 mins read
Artificial Intelligence
Artificial Intelligence Models

The Army Research Laboratory (ARL) has examined the underlying process of information exchange in reinforcement learning algorithms and identified a framework for the development of multiple artificial intelligence models that may support robot-soldier collaboration.

For the project, ARL researchers and a postdoctoral fellow at Oak Ridge Associated Universities explored algorithms within a publication period of five to six years in hopes of understanding how centralization in AI training could facilitate work on collaborative multiagent systems, the laboratory said Monday.

The team believes its findings could pave the way for further studies into teaming between autonomous technology and soldiers.

Piyush Sharma, a computer scientist at Army Combat Capabilities Development Command, said the military service sees an emerging application for multiagent systems in collaborative tactical missions as such technology becomes more prevalent in the commercial sector.

He cited Amazon's warehouse robots and Intel's drone light shows as examples of multiple systems built to work cooperatively. Sharma and his collaborators are looking to model and simulate multiagent reinforcement learning as part of theory validation and expansion efforts.

GovCon Wire, sister site of ExecutiveGov and part of the Executive Mosaic digital umbrella, hosted its AI: Innovation in National Security Forum on June 3rd. 

David Sprik, chief data officer of the Department of Defense (DOD), will serve as the forum’s keynote speaker. He will address the DOD’s defense data strategy, plans for commercial data, analytics, AI and emerging data processing technologies.

If you missed the virtual event, you can still access the OnDemand footage by visiting the GovCon Wire Events Archive.