Aurora Flight Sciences, a Boeing company, has successfully demonstrated its Fast Adaptation and Learning for Control Online (FALCON) technology, an AI-enabled control system designed to maintain safe maritime operations under challenging environmental conditions and system failures.
Developed under the Defense Advanced Research Projects Agency’s (DARPA) Learning Introspective Control (LINC) program, FALCON was created in collaboration with the MIT Aerospace Controls Laboratory and the MIT Marine Autonomy Laboratory. The machine learning-based architecture enables land, maritime, and aerial vehicles to adapt their control laws in real time, operating either as an assistant to a human operator or as the primary method of vehicle control.
A demonstration conducted in late 2025 utilized Uncrewed Surface Vessels (USVs) to simulate an underway replenishment scenario. During the exercise, a 1.5-meter-long USV was paired with a 5-meter-long vessel to maintain a consistent relative position, a task complicated by introduced hazards such as wind loading, thruster failure, and simulated Venturi effects. The team measured success based on the percentage of time the vessel remained within a defined “safe zone” and the time required to recover from induced hazards.
The results indicated a clear performance gap between manual and AI-enhanced operations. Without AI assistance, a human operator maintained the vessel within the safe zone 63% of the time. This figure rose to 82% in AI-assisted mode, where the system compensates for hazards while the human retains control. In AI-guided mode, where the operator sets parameters like speed and position for the AI to execute, the vessel remained in the safe zone 94% of the time.
Recovery metrics further highlighted the efficiency of the FALCON system. When hazards were introduced, the time required to regain control and return to the safe zone was reduced by an average of 61% under AI-guided control compared to manual operation. The development team is currently refining these algorithms in anticipation of a follow-up demonstration scheduled for the summer of 2026.
