Razvoj sistema detekcije objektov za avtonomna vozila z uporabo tehnologije LiDAR

magistrsko delo

Peter Fekonja (Author), Matej Rojc (Mentor)

Abstract

V magistrskem delu je predstavljena uporaba LiDAR sistemov in globokega učenja v kontekstu avtonomnih vozil. Delo vključuje teoretično in eksperimentalno delo. V teoretičnem delu predstavimo aktualne rešitve za razvoj LiDAR sistemov, najpogosteje uporabljene pristope za globoko učenje in metode obdelave LiDAR točkovnih oblakov z nevronskimi mrežami. Prav tako so predstavljeni aktualni senzorski sistemi na trenutni generaciji avtonomnih vozil, podatkovne baze namenjene učenju nevronskih mrež za uporabo v avtonomnih vozilih in trenutna generacija nizkocenovnih LiDAR senzorjev. V eksperimentalnem delu naloge je podrobno predstavljena zmogljivost Livox Mid-40 LiDAR sistema ter njegova uporaba v lastni rešitvi za detekcijo objektov v prometu. Podrobno je predstavljen razvoj lastne nevronske mreže kot klasifikatorja, razvoj lastnega pristopa za lokalizacijo objektov in primerjava naših rešitev z že obstoječimi pristopi. Naš pristop k lokalizaciji objektov je dosegal boljše ali primerljive rezultate z obstoječimi metodami, v kombinaciji z našim klasifikatorjem pa bistveno slabše rezultate od trenutnih enovitih modelov nevronskih mrež s prenosom znanja.

Keywords

tehnologija LiDAR;avtonomna vozila;globoko učenje;klasifikacija;lokalizacija;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2021
Typology:	2.09 - Master's Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	[P. Fekonja]
UDC:	004.032.26:004.6(043.2)
COBISS:	54864899
Views:	748
Downloads:	60
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Development of object detection system for autonomous vehicles by using LiDAR technology
Secondary abstract:	In the master's thesis we present the use of LiDAR systems and deep learning in the context of autonomous vehicles. The thesis includes theoretical and experimental parts. In the theoretical part, we present the state-of-the-art solutions for development of LiDAR systems, the most commonly used approaches to deep learning and the methods used to process LiDAR point clouds using neural networks. We also present the current sensor systems, used on the current generation of autonomous vehicles, databases targeted towards neural networks for autonomous vehicle use and the current generation of low-cost LiDAR sensors. In the experimental part, we give a detailed presentation of the capabilities of the Livox Mid-40 LiDAR system and its use in our own solution to object detection in traffic. We also show, in detail, the development of our own neural network for use as a classifier, the development of our own approach to object localization and the comparison of our solutions with existing approaches. Our object localization approach achieved similar or better results than those of existing methods, but, in conjunction with our classifier, achieved significantly worse results than current end-to-end neural network models that use transfer learning.
Secondary keywords:	LiDAR;Livox Mid-40;autonomous vehicles;deep learning;classification;localization;
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Elektrotehnika
Pages:	XVII, 228 f.
ID:	12538490