MSc Smart Manufacturing student at Nanyang Technological University with an industrial automation and system-commissioning background. I am developing practical capability in robotics software, reliable multimodal perception, and Industrial AI.
Repository | One-page research brief | Case study
- Built a controlled LiDAR-mmWave degradation evaluation around a frozen released human-activity recognition checkpoint.
- Evaluated seven lower-limb actions across 15,315 aligned target frames and a 323-condition formal matrix.
- Compared fused and unimodal behaviour and evaluated confidence calibration with reproducible tests, aggregate tables, figures, and machine audits.
- Found that quality-aware temperature scaling improved confidence reliability under the scoped protocol without changing class predictions.
Repository | Engineering case study
- Integrated Nav2 waypoint patrol, LaserScan monitoring, terminal teleoperation, and a lightweight browser dashboard in Gazebo.
- Implemented four Python application modules with approximately 2,255 lines of core code.
- Investigated command ownership, QoS compatibility, asynchronous action callbacks, and narrow-passage costmap behaviour.
- Kept claims bounded to the documented ROS 2 Humble and TurtleBot3 simulation workflow.
I am interested in connecting perception reliability to autonomous robot decision-making: deciding when a robot should continue, re-observe, replan, stop, or request human review. A related direction is digital-twin validation of recovery actions before deployment to a physical robot or production cell.
- Robotics: ROS 2, Nav2, Gazebo, RViz2, mobile-robot system integration
- Programming: Python, SQL, Shell, Linux; foundational C++
- Machine learning: PyTorch, NumPy, scikit-learn for inference, controlled evaluation, metrics, and post-hoc calibration
- Industrial automation: FANUC CNC/PMC, Siemens PLC, I/O integration, robot-CNC commissioning, Siemens NX/UG CAM
- Email: YIYANG021@e.ntu.edu.sg