This thesis focuses on multi-camera 3D object detection for smart intersections, where traffic participants are detected using multiple cameras mounted on traffic infrastructure. As automated driving continues to develop, reliability in complex scenarios such as large intersections remains a challenge. The research project VALISENS aims to more reliably capture the environment by utilizing sensor data from infrastructure through multiperspective sensor fusion.
In this work, various approaches to camera-based 3D object detection for automated driving are investigated, particularly using image data from multiple cameras. A comprehensive literature review is conducted to capture existing methods of camera-based 3D object detection in the context of automated driving. Selected approaches are implemented and evaluated on suitable datasets.
Depending on the scope of the work, there is the possibility to develop a new approach that will be included in the comparison. The goal is to assess the effectiveness and feasibility of various approaches to multi-camera 3D object detection for roadside perception and to explore new ways to improve the reliability and accuracy of environmental perception in the context of automated driving.
Work environment built on trust
Exciting and multifaceted projects
Inspiring culture of expertise
Personal development
Family friendliness
Diversity
Daniela
Tel. +49 821 885882-0
work@xitaso.com