Short Description:

Radar systems provide information about object distances, velocities and positions. Object detection is a safety critical task for automotive applications such as Advanced Driver Assistance Systems (ADAS) or autonomous driving. The time domain signal as received by an FMCW/CS radar system is processed according to the typical radar signal processing chain; the result is a range-Doppler Map. These data contain peaks at the observed object’s distances and velocities together with noise and clutter.

The task is to implement different object detection algorithms (baseline and state-of-the-art) and compare their performance using task-dependent evaluation metrics. You will work with real world measurements that were collected by our team in August 2019. The measurements don’t contain ground truth (= manual labels), so you might need to label some of the data in order to perform the evaluation.

Requirements:

• Good programming skills in Python
• Interest in code quality, code reusability and systematic evaluation