Certificate in Sim-to-Real Transfer, Digital Twins and Synthetic Data Generation

Course Overview This professional certificate provides practical knowledge and hands-on experience in building high-fidelity robotic simulations that transfer reliably to physical hardware. Participants will learn how to construct digital twins of robots and environments, author and validate USD assets, calibrate physics and sensor models against measured data, and apply domain randomisation so that models trained in simulation survive contact with the real world. The course also covers synthetic data generation at scale for perception training, including automatic annotation, dataset balancing, and the evaluation of synthetic-to-real performance. Through asset-building workshops, calibration exercises, randomisation studies, and a structured deployment case study, participants will develop the competencies required to treat simulation as an engineering instrument with quantified accuracy rather than an attractive visualisation. Suitable For • Simulation and Digital Twin Engineers • Robotics and Automation Engineers • Perception and Computer Vision Engineers • Machine Learning Engineers requiring training data • Manufacturing and Industry 4.0 Specialists • Test, Validation and Verification Engineers • Defence and Aerospace Simulation Teams • Autonomous Vehicle Development Engineers. Course Profile Level: Advanced Professional Development Duration: Five Days Delivery: Face-to-face, Online or Blended Approach: Digital twin construction workshops, USD asset laboratories, physics and sensor calibration exercises, domain randomisation studies, synthetic dataset generation sprints, and a deployment case study. Entry Requirements: Working knowledge of Python and general familiarity with robotic or 3D content pipelines. Prior exposure to simulation software is beneficial but not required. An RTX-class GPU is required for the laboratories.