Evaluation of End-to-End Automated Driving Models in CARLA Simulator

Masterthesis

There is a growing trend in industry and research towards end-to-end (E2E) automated driving models that attempt to solve the driving task directly using sensor data. Implementing and testing E2E models in automated driving simulators is a promising approach for overcoming the challenges of controlled development and certifiability.

Task details include

  • Literature review and requirement definition of E2E models for automated driving
  • Implementation and evaluation of the chosen E2E models in the open-source simulation environment CARLA
  • Discussion and comparative analysis of the capabilities of the models
  • Documentation and presentation of the methodology and results