Tuesday, February 14, 2023

Cloud-based matching of machine learning based object detection data

Topic and Goal of the Thesis

Highly automated vehicles use a variety of sensors and machine learning algorithms to detect objects in their environment. In order to expand their field of view and thus be able to see e.g. around corners, research is currently being conducted into how they can exchange their data with a cloud.

In this thesis, the aspect is to be investigated how the data that the cloud receives can be stored and processed in a meaningful way depending on its geographical location. This is to ensure that each object detected by multiple sensors is combined into a single data point without having to compare it with all the other objects.

Working Points

• Literature research on modern association algorithms
• Implementation of an efficient algorithm in an existing fusion function
• Evaluation of results based on simulated or real-world data

Requirements

• Good English or German language skills
• Passion for automated driving
• Experience with python or C++
• Experience with ROS is an advantage

Note: Please attach brief resume and grade summary.