Over recent years, new technologies for road safety and traffic efficiencies have been emerging in the context of Cooperative Intelligent Transport Systems (C-ITS) where the vehicles communicate with each other and/or with the road infrastructure. In C-ITS, vehicles are equipped with on-board sensors (such as radar, lidar, ultrasonic, camera and so forth) which, together with road-embedded sensors, collect information on the surrounding environment. However, this information is used not only by the vehicle itself, but it is also shared with the neighboring vehicles through Vehicle-To-Vehicle (V2V) or Vehicle-To-Infrastructure (V2I) communication links. The role of V2V and V2X is to extend the single-vehicle perception, limited by the on-board sensors' range, to the perception of a networked convoy of vehicles. A fundamental component of the vehicle's perception is positioning. Although Global Navigation Satellite System (GNSS) is used for vehicle localization and navigation, the accuracy and availability of GNSS do not meet the stringent requirements of most C-ITS applications, especially safety-related services. This thesis investigates a new cooperative positioning technique, denoted as Implicit Cooperative Positioning (ICP), where a set of non-cooperating objects, denoted as features, is passively detected and distributively localized by the vehicles. This information allows vehicles to implicitly enhance their own location accuracy with respect to the stand-alone GNSS solution. A key contribution of this thesis is the design of a methodology for association between vehicles measurements and features. We propose a solution based on the combination of particle-based Belief Propagation algorithms (BPA). As features are passive objects, a consensus-based approach is combined with the BPA for the distributed evaluation of features' states. Simulation results and realistic traffic flow simulation validate the improvement of the proposed ICP method with respect to the GNSS solution.
Negli ultimi anni, l'attenzione verso la mobilità sostenibile e il miglioramento dei trasporti ha portato all'introduzione di diverse tecnologie a supporto di sistemi di trasporto intelligenti (ITS). I veicoli sono diventati dispositivi tecnologicamente avanzati in grado di raccogliere informazioni sull'ambiente circostante, grazie ai numerosi sensori a bordo. Particolare attenzione è stata posta sui sistemi cooperativi, dove i singoli veicoli diventano parte di una rete connessa e in continua evoluzione. I veicoli sono connessi grazie alle innovative tecnologie di comunicazione interveicolare come la comunicazione diretta tra veicoli (nota come V2V) o la comunicazione tra i veicoli e l'infrastruttura stradale (nota come V2I). L'introduzione di questi sistemi avanzati di comunicazione permette la condivisione dei dati raccolti dai singoli veicoli, in questo modo si riesce ad aumentare la percezione del singolo veicolo estendendola a quella di un convoglio di veicoli. Un aspetto fondamentale per la mobilità intelligente è la localizzazione dei veicoli, in vista soprattutto dell'introduzione dei veicoli a guida autonoma. Questa tesi è incentrata proprio in quest'ottica e propone un nuovo metodo di localizzazione cooperativa. Tale metodo si basa sull'individuazione cooperativa di oggetti passivi collocati lungo la rete stradale da parte dei veicoli. Grazie alla condivisione di queste informazioni e alla comunicazione interveicolare, i veicoli sono in grado di migliorare l'accuratezza sulla loro posizione data dal sistema satellitare globale di navigazione (GNSS). Un aspetto fondamentale trattato in questa tesi è il metodo che permette ai veicoli di associare ciascuna misura individuale all'oggetto percepito. Tale problema non è trascurabile e richiede un'analisi complessa. Per risolvere questa problematica, in questa tesi proponiamo un algoritmo cooperativo basato su uno continuo scambio di messaggi tra i veicoli. I possibili impatti del metodo da noi proposto vengono analizzati sia in uno scenario da noi simulato, sia utilizzando dati realistici di traffico ottenuti con un simulatore professionale.
Precise vehicle positioning by Bayesian multi-target association in cooperative ITS
BRAMBILLA, MATTIA
2016/2017
Abstract
Over recent years, new technologies for road safety and traffic efficiencies have been emerging in the context of Cooperative Intelligent Transport Systems (C-ITS) where the vehicles communicate with each other and/or with the road infrastructure. In C-ITS, vehicles are equipped with on-board sensors (such as radar, lidar, ultrasonic, camera and so forth) which, together with road-embedded sensors, collect information on the surrounding environment. However, this information is used not only by the vehicle itself, but it is also shared with the neighboring vehicles through Vehicle-To-Vehicle (V2V) or Vehicle-To-Infrastructure (V2I) communication links. The role of V2V and V2X is to extend the single-vehicle perception, limited by the on-board sensors' range, to the perception of a networked convoy of vehicles. A fundamental component of the vehicle's perception is positioning. Although Global Navigation Satellite System (GNSS) is used for vehicle localization and navigation, the accuracy and availability of GNSS do not meet the stringent requirements of most C-ITS applications, especially safety-related services. This thesis investigates a new cooperative positioning technique, denoted as Implicit Cooperative Positioning (ICP), where a set of non-cooperating objects, denoted as features, is passively detected and distributively localized by the vehicles. This information allows vehicles to implicitly enhance their own location accuracy with respect to the stand-alone GNSS solution. A key contribution of this thesis is the design of a methodology for association between vehicles measurements and features. We propose a solution based on the combination of particle-based Belief Propagation algorithms (BPA). As features are passive objects, a consensus-based approach is combined with the BPA for the distributed evaluation of features' states. Simulation results and realistic traffic flow simulation validate the improvement of the proposed ICP method with respect to the GNSS solution.| File | Dimensione | Formato | |
|---|---|---|---|
|
2017_10_Brambilla.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
18.25 MB
Formato
Adobe PDF
|
18.25 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/135921