Multi-Layered Local Dynamic Map for a Connected and Automated in-Vehicle System

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

Sähkötekniikan korkeakoulu | Master's thesis

Date

2022-08-22

Department

Major/Subject

Autonomous Systems

Mcode

ELEC3055

Degree programme

Master's Programme in ICT Innovation

Language

en

Pages

62+8

Series

Abstract

Currently, the field of automated driving is receiving great attention from both large corporations and the public. Researchers are hard at work to develop fully autonomous vehicles to decrease accidents, emissions, and congestion. Although automated driving is composed of many complex and important parts, this thesis focuses on two that have lately received great attention: Local Dynamic Map and Cooperative Perception. To support current research, the aim of this thesis is to design and implement a multi-layered Local Dynamic Map that extends the standard four-layer structure by adding two additional layers. The fifth layer maps non-connected entities, such as legacy vehicles, and the sixth maps their predicted future state. The proposed solution is shown to be capable of augmenting a \sota sensor system and enable the development of efficient and accurate algorithms that aim to improve automated driving. Most notably, the matching algorithm allows to uniquely identify detected entities even if the data is received from different sources or is from a previous time instant. The prediction algorithms use motion primitives and clothoids to predict the future actions of detected entities and prevent collisions. The positioning algorithm uses the absolute position and the Course Over Ground of entities to correctly position them on the map even with high uncertainties in the data. The thesis begins by introducing the topic of automated driving. Then, it describes the concepts of Local Dynamic Map and Cooperative Perception, reviewing relevant literature. Afterwards, it designs a multi-layered Local Dynamic Map on an ideal scenario with handcrafted data, defining the overall structure. It then moves on to the implementation on real-world data collected by a prototype vehicle from CRF Trento Branch---the supporting research centre for this thesis---in urban, suburban, and highway areas of Trento. Lastly, it tests and validates the developed solution on both the handcrafted and the real-world data, drawing conclusions based on the obtained results.

Description

Supervisor

Zhou, Quan

Thesis advisor

Biral, Francesco
Visintainer, Filippo

Keywords

local dynamic map, cooperative perception, collective perception, automated driving, LDM, CP

Other note

Citation