Stochastic routing optimized for autonomous driving

Samara, Adam

[img] PDF
Dissertation_Veröffentlichung.pdf - Published

Download (14MB)

URN: urn:nbn:de:bsz:180-madoc-602065
Document Type: Doctoral dissertation
Year of publication: 2021
Place of publication: Mannheim
University: Universität Mannheim
Evaluator: Göttlich, Simone
Date of oral examination: 19 July 2021
Publication language: English
Institution: School of Business Informatics and Mathematics > Scientific Computing (Göttlich 2011-)
License: CC BY 4.0 Creative Commons Attribution 4.0 International (CC BY 4.0)
Subject: 510 Mathematics
Keywords (English): vehicle routing , autonomous driving , copulas , multicriteria optimization
Abstract: In this thesis, we propose a novel algorithm for stochastic routing optimized for autonomous vehicles. The key idea of stochastic routing is to include information on travel time reliability, rather than only estimating travel time itself. Travel time reliability is of major importance for travelers and transportation managers as it simplifies decision making and schedule planning. The concept of stochastic routing is then extended to fit the specific needs for an optimal autonomous drive. In near future, when vehicles enabled with fully autonomous driving become available, the autonomous driving features will only be possible on roads that fulfil certain criteria. Thus, when searching for an optimal route for one origin-destination-pair, we are not only interested in the travel time, but also on the route’s properties concerning autonomous driving. We estimate path travel time reliability by using empirical travel time data on segment-level. For that purpose we dive into the mathematical area of probability theory. First, we measure dependence between road segments. Then we use copulas for estimating travel time distribution on path-level by including the dependence between neighbouring road segments. In order to improve efficiency, which is needed for a real-world application, we use the following hybrid approach. We take convolution, which assumes independence, and extend it to the dependent case by integrating copulas, referred to as copula-based Dependent Discrete Convolution (DDC). Based on DDC we develop a methodology for stochastic routing. We formulate a multicriteria optimization problem, in order to find a route optimized for an autonomous drive. Different approaches to obtain one optimal solution from the Pareto front are compared, and the best fitting one is selected. This framework is then combined with the stochastic routing methodology.

Dieser Eintrag ist Teil der Universitätsbibliographie.

Das Dokument wird vom Publikationsserver der Universitätsbibliothek Mannheim bereitgestellt.

Metadata export


+ Search Authors in

BASE: Samara, Adam

Google Scholar: Samara, Adam

+ Download Statistics

Downloads per month over past year

View more statistics

You have found an error? Please let us know about your desired correction here: E-Mail

Actions (login required)

Show item Show item