Traffic management and traffic information systems

Traffic state estimation and prediction

Abstract:

As a consequence of the increased congestion in major cities, the need for more accurate traffic information with better coverage is growing all over the world. In parallel with this demand, considerably more traffic data is being collected from a variety of new and existing sources. Technologies like Bluetooth and WiFi together with different GPS-devices, smart phones, cell phone data and more traditional data sources like radar and loop detectors have the potential to create a highly comprehensive traffic database. Despite the abundance of traffic data, the types and formats available do not represent an immediate means for improving overall data quality. Their temporal and spatial resolution as well as the aggregation, accuracy and precision differ substantially. Therefore, models and algorithms that enable an optimal combination of data are necessary in order to produce accurate and reliable estimates and predictions of the current and future traffic state.

The purpose of my research is to develop models that can estimate and predict the traffic state, based on the real-time traffic data available in Stockholm.

Ph D student: Andreas Allström, andreas.allstrom@liu.se

Advisors: Professor Jan Lundgren, Dr. Clas Rydergren.

Subject: Infrainformatics.

University: Linköpings university (LiU).

Partners: Trafikverket, Sweco, KTH och University of Berkeley.

 

Travel Time Distribution Estimation/Prediction Using Floating Car Data 

 

Abstract:

The application of GPS probes in traffic management is growing rapidly as the required data collection infrastructure is increasingly in place in urban areas with significant number of mobile sensors moving around covering expansive areas of the road network. Most travellers carry with them at least one device with a built-in GPS receiver. Furthermore, vehicles are becoming more and more location aware. Currently, most systems collecting floating car data (FCD) are designed to transmit the data in a limited form and relatively infrequently due to the cost of data transmission. That means the reported locations of vehicles are far apart in time and space. In order to extract traffic information from the data, it first needs to be matched to the underlying digital road network. Matching such sparse data to the network, especially in dense urban, area is challenging.
The aim of this project is to estimate/predict the status of road traffic in terms of travel time distribution using FCD. Some applications requiring travel time information are traffic monitoring and management, journey planning, fleet management, etc. To begin with FCD, A map-matching and path inference algorithm for sparse GPS probes is developed. The method is utilised in a case study in Stockholm and showed robustness and high accuracy compared to a number of other methods in the literature.
A non-parametric travel time estimation method is developed and used to process FCD from 1500 taxis in Stockholm City. The taxi data had been ignored because of its low frequency and minimal information. The proposed method showed that such sparse data can be processed and transformed into information (e.g. travel time distribution) that is suitable for traffic studies.
The project is also contributed to the implementation of an experimental ITS laboratory, called iMobility Lab since 2010. The lab is designed to explore GPS and other emerging traffic and traffic-related data for traffic monitoring and control (visit www.imobilitylab.se for more information).
Currently, The focus is on developing a travel time prediction method using historical and real-time FCD.

 

 

Ph D student:  Mahmood Rahmani,mahmoodr@kth.se.

Advisor:  Professor Haris N. Koutsopoulos.

Subject: Traffic and transport planning.

University: Royal Institute of Technology (KTH).

Partners: Trafikverket, Sweco, IBM, LiU and University of Berkeley

 

 

Calibration of Dynamic Traffic Assignment models

 

Abstract:

Calibration of Dynamic Traffic Assignment (DTA) models is essential in order to accurately simulate various traffic phenomena in a way that the model’s outputs reflect the real-world traffic conditions.

DTA models involve various inputs and parameters needed to be calibrated. A DTA model is divided into two components: the demand models, as well as the supply models.  The demand models estimate and predict the Origin-Destination (OD) trip patterns and simulate the behavior of individual drivers. Supply models capture traffic phenomena through representations of the capacities of network elements, the speed-density relationships, etc.

The scope of this research is the development of an off-line DTA model calibration methodology for simultaneous estimation of all the demand and supply inputs and parameters, in order to minimize the error between the traffic measurements and the model’s outputs.  Various solution approaches have been proposed in literature addressing the calibration of a DTA model, but this research aims at an even more accurate and efficient solution. The methodology will be generic and applicable to any DTA system, using any general traffic data. Due to the complexity and non-linearity of the calibration problem, a major issue is the selection of an appropriate optimization algorithm.

The DTA models chosen for demonstrating the methodology that will be developed are Dynameq (Dynamic Equilibrium) and Mezzo.

 

Ph D student:  Athina Tympakianaki, athina@kth.se.

Advisors:  Professor Haris N. Koutsopoulos, Dr. Wilco Burghout, Dr. Erik Jenelius.

Subject: Traffic and transport planning.

University: Royal Institute of Technology (KTH).

Partners: Trafikverket.

 

Decision-support for Real-time Railway Traffic Management during Disturbances

Abstract:

Railways are as important part of the infrastructure in Sweden as in other countries.

With the passage of time, the railway traffic networks are becoming more and more saturated. The smooth operations of railway systems are required but due to bad weather and infrastructure failures it is hard to achieve this objective, even small traffic disturbances can propagate and have severe consequences.

 

 

The Swedish Transport Administration formally known as Trafikverket is managing the network both in terms of timetabling and traffic management while the operators arrange and run the train services for passengers and freight. The different private operators apply for desirable slots in competition with each other and Trafikverket assigns slots according to predefined market-based routines. When disturbances occur, the timetable needs quickly to be re-defined to minimize the delays and the associated penalty costs for operators and infrastructure providers. However, the large number of constraints and complex infrastructure make rescheduling difficult and time consuming. Therefore, efficient re-scheduling support for the traffic managers is needed.

 

The Swedish railway industry therefore seeks decision support systems to assist dispatchers in making good re-scheduling and delay management decisions in real time. Previously, we addressed the rescheduling problem by mathematical models and optimization based algorithms which provide help to traffic managers for better decision making. The most recent developed and tested algorithm is a sequential greedy algorithm which effectively delivers good solutions within the permitted time (30 seconds). To quickly retrieve a feasible solution the algorithm performs a depth-first branch-and-bound search using an evaluation function to prioritize when conflicts arise and then branches according to a set of criteria.

 

The main focus on the research is to design and evaluate better re-scheduling algorithms using mainly two approaches: (i) improved search and optimization techniques, and (ii) exploiting parallel solutions. The evaluation metrics are mainly how close to the optimal solution our solutions are, the execution performance (i.e., how many possible alternatives are explored per time unit), and the scalability (i.e., how large problems can we address).

Ph D student: Syed Muhammad Zeeshan Iqbal, mzi@bth.se
Advisors: Professor Håkan Grahn, BTH and Johanna Törnquist Krasemann BTH/LiU.
Subject: Computer Systems Engineering
University: Blekinge Institute of Technology (BTH).
Partners: Trafikverket.

e-infrastruktur för modern trafikledning

Abstract:
I Sverige pågår för närvarande ett antal projekt som är av stort intresse för uppbyggnaden av en ny e-infrastruktur inom järnvägsområdet. Dessa projekt innebär att man kan fånga statusen på den aktivitet och trafik som pågår på järnvägsnätet på ett helt nytt sätt. Trafikledningssystemen kan förbättras genom att man exakt vet tågsammansättning samt loket och vagnarnas position och hastighet, vilket gör att trafikledningen kan bygga på en förbättrad aktuell realtidsinformation. Syftet är att studera vad som krävs för att bygga upp en e-infrastruktur utifrån dessa förutsättningar.
Ph D student: Taline Jadaan, taline@viktoria.se
Advisors: Professor Rikard Lindgren, Docent Owen Eriksson, Professor Lars Mathiassen.
Subject: Applied IT
University: IT-universitetet i Göteborg
Partners: Trafikverket, Viktoriainstitutet.