Maritime Multi-Obstacle Tracking
Real-time multi-sensor, multi-obstacle tracking for Navantia's unmanned surface vessel.
Overview
At the GRVC Robotics Lab I developed a real-time tracking system for an unmanned surface vehicle, built with Navantia, Spain's largest naval company. It fuses several sensors to keep a consistent list of obstacles around the vessel, which feeds the downstream obstacle-avoidance and navigation modules. I also handled the LiDAR deployment and encoder integration on the platform. This became my MSc thesis: "Development of a Multi-Sensor Multi-Obstacle Tracking System for Maritime Obstacle Tracking".
I also turned the core ideas into an open training workspace (ROS 2 Humble + Gazebo). A drone tracks three moving targets by fusing ADS-B messages and LiDAR with Kalman filters, with step-by-step exercises from a single track to robust multi-target association.
Highlights
- 01Multi-target Kalman filter with track management (birth, loss and re-association).
- 02LiDAR-to-track association combined with ADS-B global-to-local transforms.
- 03Robustness to dropouts with filtering and gating.
- 04Worked with real sensor data from the vessel, coordinating with the teams deploying each sensor.
- 05Validated at GRVC; training material published as a ROS 2 workspace.