ResCon's patent-pending reservoir computing algorithms allow for implementation of robust AI/ML processes on small, low-power devices such as sUAS flight controllers.
ResCon's autopilot approach delivers real-time adjustment based on advanced state estimation and machine learning, enabling maneuver precision and complexity well beyond legacy PID controllers. Pictured: Kratos XQ-58A Valkyrie, www.kratosdefense.com
ResCon's solutions have Physical Unclonable Function (PUF) technology built in, encrypting data-at-rest, securing communication links, and providing a unique identification fingerprint.
The controller technology used to complete this record-breaking flight is part of ResCon's package of capabilities. Like a baby bird, this aircraft took off without a control scheme. Instead, it took off with a set of objectives and a machine learning algorithm that enabled it to learn how to fly in real time. This approach to flight control allowed it to survive a significant damage event when a portion of the horizontal stabilizer failed.
Video credit: The Ohio State University
Reservoir computers can be implemented in layers on edge computing devices, with a legacy controller at the lowest level if desired. This simulation shows a PID-controlled system sustaining damage and becoming unstable. A reservoir computer rapidly learns the deviation and takes control of the system, after which a second reservoir computer trains and takes control of the first. Note how little training time is required.
Video credit: Dr. Daniel Canaday
Images of aircraft are for illustration purposes only and are not meant to imply a business relationship with Kratos or any other hardware manufacturer.
Copyright © 2020 ResCon Technologies, LLC - All Rights Reserved.