Creation of an adaptive rule-based model for automated vehicles

Bachelor thesis, HES-SO // University of Applied Sciences Western Switzerland, 2020

  • 1 student to supervise.
  • The student worked full-time for 3 months.
  • 1 meeting per week.

Context of the project

To date, technological advances have made it possible to automate vehicles. For example, a level 2 automated car offers partial autonomy, i.e. the system can manage the acceleration, deceleration and steering of a vehicle thanks to the information perceived by various vehicle sensors. However, vehicle automation still causes road accidents during testing. This is why the HumanTech Institute, in collaboration with He-Arc and the University of Freiburg, has launched the “Ad Vitam” project to explore new concepts of Human-Vehicle interaction to maintain driver’s situation awareness and make future driving safer. The goal of this bachelor thesis project is to implement an adaptive model that can chose the best interface to convey information to the driver in real-time, depending on the driver’s state and the driving conditions.

Tasks done by the student

  • Review of existing frameworks to build a rule-based model
  • Choice of the best frameworks based on project constraints : Durable Rules
  • Theoretical conception of rules
  • Impementation of the rule-based model in Python
  • Communication between the model and the driving simulator
  • Functional tests
    • Simulation of driver’s state (Good vs. Bad)
    • Triggering vibrations (haptic seat), lights or nothing, depending on the driving situation
    • Sending information continuously to the mobile application