ApolloCore from Tocaro Blue is PTZ camera processing software designed to add machine learning-based perception capabilities to marine Electro-Optical and Infrared (EO/IR) cameras used on Unmanned Surface Vessels (USVs).
The developer toolset bridges commercial off-the-shelf PTZ cameras and vessel autonomy systems, providing perception capabilities for autonomous navigation, surveillance and object tracking.
The software provides automatic stabilization and ranging, real-time detection and tracking, and classification across 15 marine object classes. It uses vessel attitude data to stabilize the PTZ platform and counteract vessel motion, helping keep contacts in view even at high zoom levels. Attitude information can be supplied through an onboard Inertial Measurement Unit (IMU) or NMEA 2000 satellite compass.
ApolloCore’s machine learning models are designed for marine environments and support eight vessel classes, two marker classes and five specialized object classes. When both EO and IR channels are available, detections can also be cross-validated between the two image sources.
EO/IR Object Seeking adjusts camera aim and zoom while searching for new objects and updating existing tracks, while Fix-and-Follow can maintain a sustained track on a selected vessel and reacquire it following temporary loss.
ApolloCore can also integrate with Tocaro Blue’s ProteusCore radar processing software, combining EO/IR detections with radar information. This combines camera-based classification with radar ranging and tracking within a fused maritime perception system.
ApolloCore is deployed as a complied software executable application, built specifically for your OS and architecture, and runs on the vessel’s existing GPU hardware.
