Tocaro Blue’s ProteusCore™ Advances Radar Detection, Classification & Tracking

Tocaro Blue’s ProteusCore™ is a radar processing software platform that applies machine learning to improve the interpretation of maritime radar data.

The ProteusCore configuration application allows users to view the machine learning pipeline and configure radar and NMEA 2000 settings.

Built on a proprietary dataset of more than 3,000,000 radar frames collected and analyzed since 2019, it enables detection, classification, and tracking using off-the-shelf radar systems. The software generates outputs based on detected and tracked objects within the radar field of view.

Converting Radar Data into Structured Outputs

ProteusCore processes radar data to detect and track objects, maintaining a continuously updated list of targets for all objects around the vessel in real time. For each tracked object, the system reports position, velocity, closest point of approach, and predicted trajectory.

These structured outputs can be provided directly to external systems, including autonomous vessel control systems, where they are used to inform navigation, collision detection, and avoidance.

Automated Radar Optimization

A patent-pending Autofocus feature adjusts radar gain and range settings, replacing the need for frequent manual tuning. Autofocus modifies these settings based on environmental conditions, sea state, and proximity to land. This process ensures that your Radar adapts to different sea-states and other conditions, making the best use of the Radar’s capabilities.

Radar-Based Classification & Multi-Target Tracking

Following detection, ProteusCore applies proprietary machine learning models to analyze radar imagery and determine the class of detected objects. The system identifies small vessels, large vessels, markers, buoys, land, and wake.

The software tracks hundreds of objects simultaneously, automatically acquiring, tracking, and discarding trackers as objects move in and out of view. This classification provides additional context for each tracked object and is included in the real-time target list shared with onboard or external systems.

Radar Data for External & Unmanned Systems

Seaward Automation UI using Proteuscore Radar Processing as a perception sensor to inform autopilot systems about potential collisions. 

ProteusCore provides outputs that can be used in unmanned applications and Advanced Driver Assistance Systems (ADAS). Developers can ingest radar-classified tracks for collision detection and avoidance, allowing smart autopilot systems to make throttle and steering adjustments based on tracked targets. These outputs can be sent directly to connected systems to provide alarms and alerts to the captain.

Machine Vision Camera & Radar Fusion

As an addition to camera systems, ProteusCore offers improved distance measurement and enhanced target tracking performance. Target tracking data can inform where to point a camera and provide 360-degree awareness.

Deployment & Cost Efficiency

ProteusCore operates using CPU-based processing and does not require specialized hardware such as EO/IR camera or LiDAR-based solutions. In most cases, it can run on existing hardware and be deployed using the ProteusCore API. Tocaro Blue also offers hardware devices for ProteusCore deployments.

Applications Across the Marine Industry

ProteusCore is used across a range of environments and marine applications, including cruisers, fishing boats, workboats, and uncrewed surface vessels. The generated target list supports the monitoring of surrounding vessel activity and may also be applied in stationary installations for coastal surveillance and monitoring.

For more information, visit Tocaro Blue’s website.