Extended Object Estimation with Machine and Deep Learning Methods

Machine Learning and Deep Learning Methods grow more and more prevalent in today's scientific research, in almost all disciplines. Autonomous driving and advanced driver assistant systems (ADAS) are some of the newer areas, where these methods are employed. In current research, they're mostly used to recognize different objects in camera images, such as persons, other cars, signs, and so on, depending on exorbitant manual labour to categorize types of images

A new and exciting area of research is the investigation of high resolution radar data to detect the position of objects like cars, persons, bikes, or more. In this work, we want to focus on finding and verifying methods to extract information about extended objects in automotive scenarios, using methods of Machine or Deep Learning. These are properties like position, length, width, orientation or motion, but also the possibility to use machine/deep learning for clustering target points for later evaluation. For this, high resolution 3D radar data (range, angle and velocity) will be used.