YAPE (Your Autonomous Pony Express) is the project of a self-driving robot designed for the parcel delivery in urban areas, which must therefore be able to move in total autonomy along pedestrian routes and deal with any type of obstacle or unexpected situation that occurs in that environment. A fundamental requirement for autonomous navigation is the perception of the surrounding environment, especially in a complex and unpredictable context such as the urban one. Of particular interest are the situations in which it is required to leave the purely pedestrian route and cross the road, especially if there is no traffic light regulation, since this raises the problem of interacting with the vehicles moving on the road. The objective of the thesis is to develop a system capable of identifying vehicles in motion, tracking their position and estimating their speed, so as to have the information needed to decide when it is possible to approach the crossing safely, using the data provided by a LiDAR sensor. The results obtained with the proposed method have been evaluated on real data with application to a crossing located at a road junction, and the performance achieved have been evaluated for different types of targets (cars, motorcycles and pedestrians) and driving conditions (rectilinear motion, cornering and braking). It was concluded that the moving elements are correctly distinguished from the static ones and their position is tracked even in the presence of temporary occlusions of the field of view; the targets are also maintained and not confused with static elements even if they slow down until they stop. The obtained speed estimate, even if not precise, is useful to provide qualitative information on the nature of the motion of the tracked targets.
YAPE (Your Autonomous Pony Express) è il progetto di un robot a guida autonoma pensato per la consegna di pacchi a domicilio in ambito urbano, il quale deve quindi essere in grado di muoversi in totale indipendenza lungo percorsi pedonali ed affrontare qualsiasi tipo di ostacolo o situazione imprevista che si presenti in tale ambiente. Un requisito fondamentale per la navigazione autonoma è la percezione dell'ambiente circostante, specialmente in un contesto complesso e imprevedibile come quello urbano. Di particolare interesse sono le situazioni in cui è richiesto abbandonare il percorso puramente pedonale e percorrere un attraversamento stradale, specialmente se in assenza di regolamentazione semaforica, dove si pone il problema dell'interazione con i veicoli in movimento sulla strada. L'obiettivo della tesi consiste nello sviluppo un sistema in grado di individuare i veicoli in movimento, tracciarne la posizione e stimarne la velocità, in modo da disporre delle informazioni necessarie per decidere quando è possibile affrontare l'attraversamento modo sicuro, sfruttando i dati forniti da un sensore LiDAR. I risultati ottenuti con il metodo proposto sono stati valutati su dati reali con applicazione ad un attraversamento situato ad un incrocio stradale, e le prestazioni raggiunte sono state valutate per diversi tipi di bersagli (auto, moto e pedoni) e di condizioni di marcia (moto rettilineo, in curva e in frenata). Si è concluso che gli elementi in movimento vengono correttamente distinti da quelli statici e la loro posizione viene tracciata anche in presenza di temporanee occlusioni del campo visivo; i bersagli vengono inoltre mantenuti e non equivocati come statici anche nel caso in cui rallentino fino a fermarsi. La stima di velocità ottenuta, seppure non puntuale, è utile per fornire un'informazione qualitativa sulla natura del moto dei bersagli tracciati.
Analisi e sviluppo di un sistema di individuazione e tracciamento di ostacoli dinamici basato su LiDAR per l'attraversamento pedonale di un robot autonomo
MIGOTTO, ANDREA
2018/2019
Abstract
YAPE (Your Autonomous Pony Express) is the project of a self-driving robot designed for the parcel delivery in urban areas, which must therefore be able to move in total autonomy along pedestrian routes and deal with any type of obstacle or unexpected situation that occurs in that environment. A fundamental requirement for autonomous navigation is the perception of the surrounding environment, especially in a complex and unpredictable context such as the urban one. Of particular interest are the situations in which it is required to leave the purely pedestrian route and cross the road, especially if there is no traffic light regulation, since this raises the problem of interacting with the vehicles moving on the road. The objective of the thesis is to develop a system capable of identifying vehicles in motion, tracking their position and estimating their speed, so as to have the information needed to decide when it is possible to approach the crossing safely, using the data provided by a LiDAR sensor. The results obtained with the proposed method have been evaluated on real data with application to a crossing located at a road junction, and the performance achieved have been evaluated for different types of targets (cars, motorcycles and pedestrians) and driving conditions (rectilinear motion, cornering and braking). It was concluded that the moving elements are correctly distinguished from the static ones and their position is tracked even in the presence of temporary occlusions of the field of view; the targets are also maintained and not confused with static elements even if they slow down until they stop. The obtained speed estimate, even if not precise, is useful to provide qualitative information on the nature of the motion of the tracked targets.File | Dimensione | Formato | |
---|---|---|---|
2019_07_Migotto.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
2.28 MB
Formato
Adobe PDF
|
2.28 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/10589/148610