CRA employs deep learning algorithms to extend detection capabilities in TOPGUN effortNews
January 10, 2019
CAMBRIDGE, Mass. The Office of Naval Research (ONR) tasked Charles River Analytics Inc. to build a maritime defense tool for the Topside Optical Processing for Global Unmanned Navy (TOPGUN) effort. The 18-month program is valued at $1.6 million.
Under the program, Charles River will support the TOPGUN effort to continue autonomy development, extending detection and classification capabilities, with efficient deep learning algorithms to detect more distant targets, recognize a wider range of ship types, and demonstrate performance on the water. TOPGUN software will be hardened to meet information assurance requirements and tested at sea.
“We’re excited about the TOPGUN effort because it allows us to refine our already impressive CROWSNEST ship detection and classification results and demonstrate these capabilities on a truly innovative platform,” says Ross Eaton, Principal Scientist at Charles River Analytics. “This demonstration will improve safety and reliability by supporting small craft detection and COLREGS compliance, necessary steps on the path to enabling the full range of autonomous capabilities. Because TOPGUN is camera-agnostic, we can apply this same technology to new programs and new platforms, providing autonomous ship detection and classification in timeframes that are unmatched.”
“TOPGUN is just one of many ongoing Charles River efforts across the unmanned systems domain where we bring extensive experience and capabilities to the challenges of deploying truly autonomous robotic platforms. We apply our state-of-the-art sensor processing and AI technologies to on-board autonomy that enables unmanned systems to ‘see,’ understand, and respond independently to the operating environment,” commented Rich Wronski, Vice President of Charles River’s Sensing, Perception, and Applied Robotics division.