Abstract

Urban Freight Distribution (UFD) represents a very important activity in terms of cities economy, but can also be a problem for the daily life of citizens, due to traffic, noise and environmental contamination. Therefore UFD activities must be managed cautiously, to avoid or minimise external negative effects. In this context, this research aims to evaluate existing UFD measures, ranking the m from the best to the worst one, according to experts’ opinions. The proposed evaluation process is flexible in order to allow evaluating future UFD measures that may arise.

Abstract

Digital technologies have increased the influence of technology in business, even changing busi- ness models and strategies of organisations. This influence, called Digital Transformation of Business (DT), happens when there is an increase of the number of digital connections, infor- mation and interactions. This phenomena has been defined as Digital Density (DD) and aims to provide an assessment of the digitalization status of an organization. With the concept of DD we propose the DD Framework, that has two parts: qualitative and quantitative. In order to support both parts we simulate the effect that a platform for smartphones called Waze has on the traffic flow of a city. For the qualitative side, we redefine the concept of DD and with Business Model CANVAS we compare the changes introduced by Waze in the business model. The quantitative part of the simulation is performed with MATLAB and its goal is to under- stand how an agent chooses the path to drive from point A to point B. As expected, we prove that with more information agents can estimate better the costs of the different paths. Despite, even if our simulation seems to prove that the more information the agents have the better the system works, we introduce an analogy with the Braess Paradox to find potential problems and provide solutions for them. We conclude with a proposal of future work.

Abstract

The main objective of this thesis is to establish a bicycle system in Tallinn. It will explain the whole process for establishing the system. The first part consists in analyze the different bike sharing systems in Europe and with the obtained data it is decided which si the most suitable operating system for Tallinn. This analysis is complemented with calculations and graphs that are going to help us to determinate which parameters have to has the city of Tallinn for has succeed in the installation. The second part is based on the location criteria for the stations of the system. Using location models, programming and solving it all an optimal solution will be found.

Abstract

In today’s urban transportation networks, traffic is regulated by traffic signals which control vehicle flows in road intersections and other road spaces. Generally, controllers follow a fixed-cycle schedule which is previously designed. Under heavy or changing conditions, these standard methods may prove insufficient, and it is where adaptive traffic signaling at intersections can be the solution. This thesis examines a part of Helsinki’s city center network to verify theories of traffic flow and tests the performance of the Max-Weight algorithm under different scenarios. Traffic simulations are accomplished using PTV Vissim, which is a microscopic traffic simulator, whereas MATLAB R2018b executes the algorithms, the filter, and plots the results. The analysis led to the following conclusions: the Max-Weight algorithm control outperforms the traditional fixed and cyclic control under rush hour situation in Helsinki. Nevertheless, under noisy measurements, the tested filtered Max-Weight algorithm control did not make such a difference, compared with the non-filtered noisy control. Further investigations are proposed referring to the particle filter.

Abstract

This project studies the aggravating problem of road traffic and congestion, and some of its solutions: autonomous cars and traffic simulation software. Unreal Engine is presented as a new option for simulating a traffic environment. An application is programmed that allows the user to create and edit roadtracks by clicking and dragging in the editor, along with autonomous vehicle models to populate them with. The simulator is tested using different scenarios and obtaining relevant data. Finally, the obtained conclusions are mentioned, along with ideas and guidelines for future developmen

Abstract

Maritime transport covers the immense majority of the world’s international trade, particularly for Spain, where it represents a 74% of the total transported goods. Most of this transport is made via containers, because of its versatility. Its drawback is that very specific infrastructure and large spaces are needed for its management, and these are not fast to build nor cheap. In addition, the inability to manage containers if the infrastructure is under work or has reached his maximum capacity may translate into important economic losses. To prevent this from happening it is mandatory to plan in advance, and that means accept some predictions of the future to act accordingly. There is a wide range of accepted forecasting methods, because none is good in every situation, and predicting container throughput is not an exception. In this project, we will discuss the validity of Markov Chains when it comes to forecast maritime goods exchange, more precisely the total annual volume of TEUs. As real cases, we will take the data of 4 important Spanish ports (Algeciras, Barcelona, Tarragona and Valencia) from 1973 to 2015 and use it to create a prediction model. We will use the Markov property to create predictions and carry out Monte Carlo simulations to sample them. Over the sections we will change the input data and the parameters of the model to better understand the behavior of Markov chains when applied in such area. To have a reference, we will perform a witness forecast by means of a linear regression. An accuracy measure is planned for each of the models to assess a comparison between them, and find which one is more precise, and thus more valid. El transporte marítimo abarca la inmensa mayoría del comercio internacional en todo el mundo, y para España en particular, donde representa el 74% del total transportado. La mayor parte de este transporte se realiza en contenedores, por lo versátiles que resultan. Lo malo de este sistema es que se necesita una infraestructura específica y mucho espacio para poder procesarlos, lo cual tarda tiempo en construirse e implica grandes inversiones. Además, el hecho de no poder recibir contenedores por remodelaciones en la infraestructura o esta ha llegado a su máximo puede traducirse en cuantiosas pérdidas económicas. Para evitar que esto ocurra es de vital importancia planear de antemano, lo cual implica aceptar una predicción y actuar en consecuencia. Existe un amplio abanico de métodos predictivos, esto se debe a que ninguno de ellos es válido en todas las situaciones posibles, y en el ámbito del transporte marítimo no es una excepción. En este Proyecto discutiremos la validez de las cadenas de Márkov a la hora de prever el intercambio de mercancías por vía marítima, más concretamente el número total de TEUs por año. Como estudiar casos reales tomaremos los datos de 4 importantes puertos españoles (Algeciras, Barcelona, Tarragona and Valencia) desde 1973 hasta 2015 y los utilizaremos para probar el modelo. Utilizaremos la propiedad de Márkov para crear las predicciones y llevaremos a cabo simulaciones Monte Carlo con el fin de muestrearlas. A través de los diferentes apartados iremos cambiando los datos y los parámetros del modelo para entender mejor el comportamiento de las cadenas de Márkov cuando se aplican en esta área en particular. Para tener una referencia, utilizaremos como experimento testigo una predicción realizada con una regresión lineal. Está planeada también una medida de la precisión de cada uno de los modelos, para poder así compararlos y evaluar cuál es el que mejor se ajusta a la realidad, y por lo tanto es el más valido.

Abstract

Treball fi de màster de: Master's Degree in Data Science Directors: Hrvoje Stojic, Anestis Papanikolaou In this thesis we develop a traffic light control agent that can manage traffic lights with the objective to reduce traffic jams, trip time and other traffic metrics in a given network using reinforcement learning. To this end, we implement a Double Deep Q-Network algorithm and test its performance in controlling traffic lights on a ’small’ and a ’large’ traffic junction. We find that this algorithm beats a fixed traffic light phase program when traffic demand fluctuates, as it is capable of reacting to real-time traffic situations. The algorithm can be scaled up and holds promise to also perform well in controlling larger transport networks. En aquest treball de final de màster es desenvolupa un algorisme d'aprenentatge reforçat pel control de semàfors amb l'objectiu de reduir temps de trajecte i retencions. Específicament, s'ha implementat l'algorisme Double Deep Q-Network i s'ha comprovat la seva eficàcia comparant-lo amb escenaris realistes de control d'una intersecció simple i d'una complexa. S'ha demostrat que aquest algorisme es comporta millor que l'escenari real en el qual el canvi de fase es duu a terme amb intervals de temps fixes. Els resultats indiquen que aquesta tècnica és capaç d'adaptar-se a les situacions de trànsit canviants i per tant obtenir millor resultats que l'escenari real. L'algorisme pot ser adaptat per controlar xarxes de trànsit més grans.