Karthikeyan Chandrasekaran

After completing the module, students are able to:

- explain the mathematical and statistical foundations of uncertainty modeling, distinguish between epistemic and aleatoric uncertainty, and contrast modular and end-to-end AD-stack architectures;
- evaluate methods for quantifying and calibrating uncertainty in perception, identify causes of miscalibration, and select appropriate improvement methods;
- analyze how cooperative V2X systems both reduce and introduce uncertainty, and classify concepts for representing, communicating, and propagating uncertainty in cooperative perception architectures;
- describe human behavior and cognitive models of drivers and VRUs and incorporate them into motion prediction;
- formulate probabilistic models for the motion prediction of traffic participants, trace the propagation of uncertainties along the prediction-to-planning chain mathematically, and apply uncertainty-aware planning methods (e.g., RRT variants);
- assess methods for scenario-based validation, runtime monitoring, and the safe integration of automated driving functions under uncertainty, and judge their suitability against relevance and criticality criteria.