Jamie Bennet
General Atomics‘ aeronautical systems business has accomplished a completely autonomous flight of the Avenger MQ-20A, an experimental unmanned aircraft system designed for military use.
The Sept. 12 test flight employed an artificially intelligent pilot that navigated the vehicle for nearly 30 minutes and aimed to validate a company-built reinforcement learning architecture in an operationally relevant environment, General Atomics Aeronautical Systems Inc. said Monday.
GA-ASI’s team used RL agents to fly “chase and avoid behavior,” which involved using live fused tracks to make real-time flight path updates to evade enemies.
The live fused tracks utilized TacIRST, an infrared search and track sensor network from Lockheed Martin.
The team also applied government standards, notably the Open Mission Systems messaging protocol and an autonomy engine based on the Collaborative Operations in Denied Environment program.
MQ-20A’s flight is part of an autonomous mission series the company is performing to demonstrate AI and machine learning applications in UAS.
