Master Thesis Development of an ML-Based Post-Processing Scene Classification for ADAS Development Using Continuously Evaluated Data

In recent years, Advanced Driver Assistance Systems (ADAS) such as Adaptive Cruise Control (ACC) and lane departure warning have become increasingly popular. However, these systems still have room for improvement in certain scenarios, which requires a systematic search for such scenes during the development phase. Unfortunately, this process requires expensive and time-consuming manual expert labelling to cluster and label these effects. To address this problem, in your Master thesis you will design a machine learning algorithm using continuous-valued input data from radar and camera to classify faulty scenarios in post-processing. Furthermore, you will test and optimize the algorithm using real data. Last but not least, you will integrate the algorithm into an existing validation framework.