- Design and manage the field data collection pipeline - including body-worn cameras, fixed-position gesture capture cameras, and pole-mounted lane closure cameras deployed at active road worksites - Annotate, label, and structure video datasets capturing vehicle behaviour, ASO gestures, and traffic compliance events - Train and optimise computer vision models (YOLOv8 or equivalent) for vehicle detection, trajectory tracking, and driver compliance classification in outdoor road worksite environments - Calculate traffic safety metrics including Post-Encroachment Time (PET) to identify and prioritise near-miss events for model training - Build and refine logit or ML-based conflict severity models to support edge case identification in the training dataset - Export trained models to TensorRT and deploy on NVIDIA Jetson Orin NX hardware - Develop ROS2 integration for deploying the perception pipeline onto a humanoid robot platform - Build multi-robot coordination logic using ROS2 pub/sub a...