Friday, February 02, 2024
Machine Learning Based Object Detection in 4D-Radar Point Clouds
Topic and Goal of the Thesis
Advanced Driver Assistance Systems (ADAS) and automated driving technologies rely on sophisticated automotive perception systems. Radar sensors are a fundamental component of these systems, supporting functionalities like Adaptive Cruise Control (ACC) and collision avoidance. Recent advancements in radar technology, such as 4D Radar, offer the potential for a more comprehensive perception, enabling more accurate and robust driving functions.
Machine Learning-based point cloud object detection is a tool for environmental perception in vehicles that has been extensively researched for LiDAR sensors [ZAM21]. However, for radar data, and especially modern 4D radar data, this method is significantly less explored. This thesis shall investigate the potential of machine learning-based object detection in radar point clouds.
Working Points
- Comprehensive literature research on object detection in radar point clouds and existing approaches in this domain.
- Identification of suitable machine learning algorithms.
- Development of a machine learning-based approach for 4D radar object detection and implementation into the existing framework at the institute.
- Evaluation of the implemented algorithm using metrics based on a sample dataset.
Requirements
- Good English or German language skills
- Reliability, commitment, and enjoyment of working independently
- Programming skills, preferably in Python
- Experience in machine learning
Note: Please attach brief resume and grade summary.
Contact
Lukas Ostendorf M. Sc.
+49 241 80 25624
Email
Type of work
Bachelorarbeit, Masterarbeit
Start
earliest date possible
Prior knowledge
Programming skills in Python, experience in machine learning
Language
Deutsch, Englisch
Research area
Fahrzeugintelligenz & Automatisiertes Fahren