Yohan Le Gars

• Designed and implemented multi-object tracking and multi-sensor fusion pipelines using AIS, LiDAR, and radar data.
• Improved the camera pipeline using GStreamer and integrated D3 camera systems into the perception stack.
• Optimised Ouster LiDAR pipeline, performed LiDAR tuning, point-cloud noise filtering, and third-party compression using Cloudini.
• Integrated robotics and sensor tooling, including Foxglove for visualization/debugging and IO-Link IFM drivers for industrial sensor communication.
• Developed ground station software features, including Qt-based radar plot, user interface components, and alerting systems.
• Built target-scene logic and user interface components in Godot.
• Supported deployment and development workflows using CI/CD and Docker.

Master’s Thesis — Multi-Sensor Fusion and Multi-Object Tracking

• Developed ROS 2 C++ LiDAR-AIS fusion package for multi-object tracking, enhancing tracking accuracy through JPDA-based IMM-EKF framework.
• Customized nuScenes SDK to retrieve and analyse tracking performance metrics, improving data-driven insights for performance evaluation.
• Generated CoppeliaSim-based simulated dataset to validate sensor fusion and tracking performance, supporting robust algorithm testing.

• Recorded scenes in Isaac sim to evaluate performance of spatial mapping algorithms.
• Automated labelling for semantic segmentation.
• Training and testing SimpleRecon 3D reconstruction model.