LiDAR is all you need – 3D Object detection for autonomous Driving

Institut
Lehrstuhl für Fahrzeugtechnik
Typ
Semesterarbeit /
Inhalt
experimentell /  
Beschreibung

Lidar has emerged as the solution for the issues that camera cannot solve in the topic of au-
tonomous driving. Depth estimation and lack of spatial information are problems of the past,
thanks to lidar sensors. Therefore, if Waymo can successfully use LiDAR to detect 3D ob-
jects in real-time, so does EDGAR. The goal of this project is to implement a single and
multi-LiDAR object detection module to drive around Munich.
As part of your student thesis, you will expand the existing EDGAR software stack by incor-
porating a cutting-edge 3D LiDAR detection module. To accomplish this, your work will in-
volve researching suitable architectural choices based on criteria such as execution time,
computational demands, and performance. Subsequently, you will implement and train these
architectures using established datasets, followed by a comprehensive evaluation of your re-
sults. This evaluation will encompass quantitative assessments using validation data, as well
as qualitative evaluations through real-world vehicle testing.


Work packages:
• Literature review of 3D object detection
• Autonomous driving dataset review, collection or generation
• Object detector training and evaluation and module implementation in Pytorch
• Implementation of the module as a ROS2 node inside the Autoware stack
• Test of the module on the EDGAR vehicle

 

Voraussetzungen
  • Requirements:
  • Programming experience in Python or C++
  • Experience with Pytorch/Tensorflow
  • Knowledge of computer vision,
  • Desired: Experience with ROS or ROS2
Software packages
Python, Pytorch, Tensorflow, ROS2
Tags
FTM Studienarbeit, FTM AV, FTM AV Perception, FTM Rivera, FTM Informatik
Möglicher Beginn
sofort
Kontakt
Esteban Rivera, M.Sc.
Raum: MW 3508
esteban.riveratum.de
Ausschreibung