Predictive diagnostic of Modular Autonomous Electric Vehicles

  • Subject:Machine learning, Autonomous Electric Vehicle, Predictive diagnostics
  • Type:Masterarbeit
  • Date:ab 03 / 2024
  • Tutor:

    M. Sc. Veljko Vucinic

    M. Sc. David Kraus

  • Zusatzfeld:

    Please note: The Master's thesis will be written exclusively in English.

Predictive diagnostics of Modular Autonomous Electric Vehicles

Demonstrator U-Shift II (Demonstrator)

Context

Following trends of electrification and automation of vehicles, the field of diagnostics rises concerns. Electrical vehicle powertrains do not produce CO2 emission, therefore the standard on-board diagnostic systems are not obligatory for these vehicles. Regardless, there is a demand for diagnostic system in order to increase durability and reliability of an electric vehicle. During the development of the prototypical fully automated vehicle U-Shift, the absence of human driver raised critical future issues in autonomous vehicle diagnostics (i.e. driver can react to unknown in vehicle vibrations, smell, etc.). Implementing predictive diagnostics on critical autonomous electric vehicle systems like Battery, Battery Management System, electromotor drives, etc. can address these problems and therefore shape the future of autonomous electric vehicle diagnostics.

Tasks

  • Overview of the state of the art of electrical and autonomous vehicle diagnostics
  • Investigating the best approaches to predictive diagnostics of modular autonomous electric vehicle
  • Development of ML models based on real data from critical U-Shift systems (Battery, BMS, Electromotor drive, Power electronics)
  • Evaluation of predictive system developed
  • Deployment of the best ML predictive diagnostic model on the U-Shift prototype

Requirements

  • Interest in the field of AI/ML systems
  • Basic knowledge and experience in data processing
  • Programming skills (Python/C++/Matlab)