2025

• Twin-embedded MPC for longitudinal/lateral vehicle control
  
  Development of a high-fidelity, physics-based vehicle dynamics simulator in Modelica/Dymola, integrated with a digital twin for predictive control tasks.
  
  Key question: Can we reliably initialize a simulator using observed vehicle states?
  
  Tools & Skills: Modelica/Dymola, Python, Matlab, Simulink
  
  
  • Development of physics-based digital twin
  • Integration with Model Predictive Control frameworks
  • Development of black-box non-linear controller (e.g., Bayesian Optimization [BO])
  • Testing in simulation
  

  Cruise control application – Bayesian-based throttle controller.

• Man-in-the-Middle (MiTM) Attack Detection [Cybersecurity/ML]
  
  Development of detection algorithms against MiTM attacks on control systems, focusing on resilient predictive controllers.
  
  Tools & Skills: Kali Linux, Python, PyTorch/TensorFlow, Wireshark, Matlab, Simulink
  
  
  • Experimental deployment on actual networks
  • In-depth study of industrial communication protocols
  • Development of ML-based intrusion detection algorithms
  • Design and testing of resilient predictive controllers under attack
  

  Man-in-the-Middle - Let's detect it!