Autonomous Systems & Planning

Multi-Step Deep Koopman for Vehicle Control in Frenet Frame

Embedded a Python deep-Koopman vehicle model in MATLAB/Simulink for trajectory-tracking MPC of a high-fidelity CarSim vehicle.

Implementation Cross-language Python–Simulink MPC integration with the CarSim-modeled vehicle; the work was presented at IROS 2025.

  • Implemented cross-language integration by embedding Python in MATLAB/Simulink for a trajectory tracking MPC for high fidelity CarSim-modeled C-Class Hatchback vehicle.
Comparison between MPC performance with LTI model vs MDK model and the reference trajectory.
  • Presented at IROS 2025.

Skills

  • Python
  • MATLAB/Simulink
  • MPC Design
  • Koopman Operator
  • Deep learning-based system identification

Related Publication

Multi-Step Deep Koopman Network (MDK-Net) for Vehicle Control in Frenet Frame

IROS 2025

View Publication Paper