Skip to content Go to main navigation Go to language selector
Saab Global
reflex-ai-for-mobile.png.jpg

Saab and Purdue University Team Up on DARPA Program to Advance Adaptive Controls for Uncrewed Sea Vessels

3 min read

Phase 1 of DARPA’s LINC program aims to improve sea vessel capabilities in instances of compromised system dynamics and control authority.

For autonomous vehicles, it’s not enough to build a system capable of following mission parameters in ideal situations. In these systems, it’s equally important to create mechanisms that enable corrective options if a vessel deviates from its planned behavior or finds itself in less than perfect conditions. To address this challenge, Saab, Inc. and Purdue University are working with the Defense Advanced Research Projects Agency (DARPA)  as part of their Learning Introspective Control (LINC) program, and the solutions being developed are already changing the game. 

Under the LINC program, Saab and Purdue have developed Explainable Reflexive Control (RefleXAI), a cutting-edge system that uses machine learning-based technologies to enable self-correcting control for uncrewed sea vessels (USVs). RefleXAI enables the USV’s onboard sensors to determine whether the system has strayed from normal or planned behavior and analyze the impact of that change on system controls and operations. The system then updates the vessel to maximize control, minimize loss in mission objectives and maintain safe and efficient operations.

“The current approach to handling a system with compromised control authority places much of the burden on the human operator,” said Dr. Christopher Vo, deputy chief scientist within Skapa, Saab’s innovation incubator. “In the RefleXAI project, we seek to create a self-healing system that would have the ability to learn and adapt to complex and dynamically changing situations, reducing the workload of the human operator and allowing them to focus on their mission safely and effectively.”

Saab and Purdue have rigorously developed and tested RefleXAI across various scenarios, such as maintaining a consistent position relative to another vessel – a crucial element for refueling operations at sea. These tests ensure that the USV can execute such tasks successfully, even while navigating disturbances including strong wind or mechanical failures.

linc2jpg.jpg
Figure 1: The WAM-V 16 USV Testbed Platform for LINC Algorithms.

Looking ahead, Saab plans to test RefleXAI control architecture on the Enforcer 3, an autonomy platform built from a converted Combat Boat 90. Equipped with advanced navigation and communication systems, sensors and cameras, the Enforcer 3 will leverage RefleXAI to enable remote operators to control the vessel safely and reliably during the most challenging tasks.

“When Saab and Purdue University work together, we are able to accomplish incredible things. We’re excited to continue our partnership on this project, and to leverage the expertise, technology and ingenuity that both our teams bring to the table. Together, we can advance critical system capabilities for aquatic vehicles, which play a significant role in our mission to keep people and society safe.” said Michael Brasseur, vp, chief strategy officer and gm of Skapa.