In the last few years huge strides have been done by car industry in the field of autonomous vehicles. Currently, several systems of driving assistance (ADAS), are increasingly implemented on vehicles, in order to simplify and help people during the driving. Moreover, several industries and universities are developing new prototypes to achieve a full automated vehicle able to move in every environment, whether urban or motorway. It has to be noticed that in this process of automation, several aspects need to be taken in consideration. In particular, much attention is put on sensors, to provide reliable information of the surrounding environment and on path planning algorithms to drive the vehicle. Our thesis is focused on these two topics to develop an algorithm to drive an autonomous vehicle in urban environment. In particular, we focused the attention on the reconstruction of the static infrastructure in front the vehicle such as shape of the road ahead and presence of junctions or roundabouts, adopting an innovative system based on electronic maps called E-Horizon. Using the trajectory generated with the E-Horizon system as a reference, we implemented a path planning algorithm based on Model Predictive Control to conduct the vehicle. The novelty presented in this algorithm concerns the implementation of new features that make this algorithm able, not only to avoid the obstacles, but even to consider the limits of the road. These features are fundamental to move the vehicle in urban environment in safety. Moreover, the output has been designed in order to avoid instability and conflict with the controller of the actuators. This work was developed in the Biomechanics section of the 3ME Department of Delft University of Technology, during a period between 22 November 2015 and 31 May 2016.
Negli ultimi anni, l’industria automobilistica ha compiuto molti passi avanti nel campo dei veicoli autonomi. Attualmente diversi sistemi di ausilio alla guida, indicati con l’acronimo inglese ADAS (Advanced Driver Assistance Systems) sono sempre più impiegati a bordo delle automobili, allo scopo di semplificare e supportare gli automobilisti durante la giuda. Inoltre, diversi costruttori e università stanno sempre più sviluppando nuovi prototipi di veicoli autonomi in grado di muoversi in diversi scenari, sia urbani che autostradali. E' da notare che in questo processo di automazione, diversi aspetti tecnici devono essere considerati. In particolare, molta attenzione è stata posta sui sensori, allo scopo di fornire informazioni affidabili riguardo l’ambiente circostante, e sugli algoritmi di pianificazione della traiettoria. La nostra tesi è incentrata su questi due argomenti, allo scopo di sviluppare un algoritmo di pianificazione della traiettoria per un veicolo autonomo in ambiente urbano. In particolare ci siamo occupati di ricostruire l’ambiente circostante il veicolo, in termini di forma della strada con eventuale presenza di incroci e rotonde, avvalendoci di un innovativo sistema basato su mappe elettroniche chiamato E-Horizon. Usando la traiettoria generata con il sistema E-Horizon come riferimento, abbiamo implementato un algoritmo di pianificazione della traiettoria basato sulla logica chiamata Model Predictive Control. L’innovazione presentata in questo algoritmo concerne principalmente l’introduzione di nuove caratteristiche in grado di non solo di evitare ostacoli ma anche di tenere in considerazione il limite del bordo strada. Queste particolarità sono fondamentali per condurre il veicolo in modo sicuro e confortevole. Inoltre l’output di questo algoritmo è stato progettato in modo da limitare al massimo l’instabilità del controllore di basso livello degli attuatori. Questo lavoro è stato sviluppato nella sezione di Biomeccanica del Dipartimento 3ME dell’Università Tecnica di Delft, durante il periodo di tempo inziato il 22 novembre 2015 e finito il 31 maggio 2016.
Safe path plannng algorithms for autonomous vehicles in urban environment
VACCHELLI, ALBERTO;ROMANINI, LORENZO
2015/2016
Abstract
In the last few years huge strides have been done by car industry in the field of autonomous vehicles. Currently, several systems of driving assistance (ADAS), are increasingly implemented on vehicles, in order to simplify and help people during the driving. Moreover, several industries and universities are developing new prototypes to achieve a full automated vehicle able to move in every environment, whether urban or motorway. It has to be noticed that in this process of automation, several aspects need to be taken in consideration. In particular, much attention is put on sensors, to provide reliable information of the surrounding environment and on path planning algorithms to drive the vehicle. Our thesis is focused on these two topics to develop an algorithm to drive an autonomous vehicle in urban environment. In particular, we focused the attention on the reconstruction of the static infrastructure in front the vehicle such as shape of the road ahead and presence of junctions or roundabouts, adopting an innovative system based on electronic maps called E-Horizon. Using the trajectory generated with the E-Horizon system as a reference, we implemented a path planning algorithm based on Model Predictive Control to conduct the vehicle. The novelty presented in this algorithm concerns the implementation of new features that make this algorithm able, not only to avoid the obstacles, but even to consider the limits of the road. These features are fundamental to move the vehicle in urban environment in safety. Moreover, the output has been designed in order to avoid instability and conflict with the controller of the actuators. This work was developed in the Biomechanics section of the 3ME Department of Delft University of Technology, during a period between 22 November 2015 and 31 May 2016.| File | Dimensione | Formato | |
|---|---|---|---|
|
2016_09_Romanini_Vacchelli.pdf
non accessibile
Descrizione: Testo completo della tesi
Dimensione
11.88 MB
Formato
Adobe PDF
|
11.88 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/10589/124147