Project Goals 

Creating optimal traffic light schedules is of utmost importance to urban commuters, encompassing vehicle drivers, pedestrians, businesses, governing authorities, and city residents as a whole. Nonetheless, manually tackling this task proves intricate due to the multitude of constraints involving vehicle drivers, pedestrians, roads, junctions, and traffic lights that must be considered. Hence, exploring the automated generation of these traffic light working schedules and synchronization is not only scientifically intriguing but also significantly crucial in the context of urban cities.

The objective of this project is to address this challenge precisely by devising novel optimization techniques rooted in artificial intelligence and machine learning. These methods are intended to optimize traffic light scheduling for the city of Prishtina and explore cutting-edge approaches that could potentially be adapted for Prishtina's context. Our goal is to formulate hybrid optimization methods and employ machine learning-based reasoning to resolve this envisioned issue. Furthermore, we will develop a prototype system for traffic light scheduling and synchronization, which will initially cater to Prishtina's needs but will possess the flexibility for future extension for use in other cities in Kosovo or even internationally.

The project will make a valuable contribution to the research field of traffic light scheduling and synchronization, fostering collaboration between academia (University of Prishtina), industry (Comitas AG and Golden Taxi), local government (Municipality of Prishtina), and international universities (TU Wien). Furthermore, this project presents a significant opportunity for young Kosovar researchers to enhance their expertise in web technologies and gain a strong research foundation in the areas of artificial intelligence, machine learning, optimization, and automated scheduling.

In line with the call for proposals, this project falls into the category of addressing concrete community issues through collaboration between higher education institutions and SMEs. In our specific case, the project focuses on tackling the problem of traffic congestion in urban areas. As a case study, we will address the issue of automated traffic light scheduling for the city of Prishtina. Our aim is to generate improved traffic light schedules, benefiting the population of Prishtina and all commuters in various ways, including reduced commuting times, lower transportation costs, decreased air pollution within the city, and enhanced safety at traffic intersections for both vehicles and pedestrians.

The project comprises six primary activities, which are as follows:

1. Analysis of requirements and formal definition of the traffic light scheduling problem for Prishtina, involving the creation of new problem instances based on real-world requirements.

2. Investigation and comparison of state-of-the-art methods applicable to the defined problem.

3. Design and implementation of novel hybrid optimization methods tailored to the context of Prishtina.

4. Development and implementation of a prototype system for manual and automated traffic light scheduling.

5. Facilitating collaboration and networking among researchers from academia (University of Prishtina / TU Wien), industry professionals (Comitas AG), and local government authorities (Municipality of Prishtina).

6. Presentation of research findings at scientific conferences and/or peer-reviewed journals.

The project can be segmented into four key milestones, outlined as follows:

1. Formal definition of the specific traffic light scheduling problem tailored to the city of Prishtina.

2. Assessment and comparison of state-of-the-art methods found in the existing literature.

3. Design, creation, and evaluation of innovative approaches for traffic light scheduling.

4. Development and implementation of a prototype system for both manual and automated traffic light scheduling.

Project duration

July 2023 - May 2024

Team 

Universiteti i Prishtinës (University of Prishtina)

• Kadri Sylejmani (Project coordinator and Lead Researcher)

• Bujar Krasniqi (Researcher)

• Lavdim Kurtaj (Researcher)

Master students 

• Fjolla Gashi (Master thesis: Modelling and generating vehicle traffic in the City of Prishtina using real-life data)

• Atlantik Limani (Course project: Solving the Traffic Signaling Problem Using the Artificial Bee Colony Algorithm)

• Elvir Misini (Innovative project: Solving the Traffic Signaling Problem Using Iterated Local Search Metaheuristic) 

• Elon Demi (Participant in innovative project: Developing a web tool for automated traffic signal scheduling)

• Festina Qorrolli (Innovative project)

• Fisnik Spahija (Innovative project)

Bachelor students

• Erzen Krasniqi (Course project: Developing a Web-based Test Instance Generator for Traffic Light Scheduling)

• Bledian Potera (Bachelor thesis: Developing a web tool for the graphical presentation of test cases for the traffic light scheduling problem)

• Aurel Demiraj (Bachelor thesis: Developing a web application for the graphical representation of traffic lights in the city of Prishtina using JavaScript libraries)

• Eriona Mustafa (Bachelor thesis: Design and Implementation of a Simulator using the SUMO Library)

• Elona Rashica (Bachelor thesis: Developing a web tool for visualization of taxi paths in online maps)

• Blinera Hysenaj (Bachelor thesis: Visualizing urban mobility traffic data for the City of Prishtina using JavaScript libraries)

• Fisnik Mustafa (Bachelor thesis: Development of a web application for the graphical presentation of traffic light scheduling solutions)

Open projects in master studies 

• Prediction of driving duration in the City of Prishtina using Gradient Boosting Algorithms

• Analysing Bus GPS Data for Uncovering Traffic Patterns, Identifying Peak Travel Times, and Mitigating Recurring Congestion in Urban Transportation 

Open thesis in bachelor studies

• Developing a web application for aggregating GPS data of taxi vehicles in the city of Prishtina

• Data visualization and analysis of taxi services in the City of Prishtina using Python libraries

• Creating a web application for the geospatial visualization of roads and junctions within a city

• Developing a Web Application for Geospatial Visualization of Traffic Lights in the City of Prishtina

• A Simulation Study of Vehicle Traffic and Intelligent Traffic Light Control in Prishtina

• Design and Implementation of a Web-Based Traffic Simulator using the SUMO-GUI Library

Comitas AG

• Adriatik Ademi (Project coordinator for Comitas Services LLC | Software developer coach)

• Mihrije Kadriu (Software developer coach)

TU Wien: Technische Universität Wien

• Nysret Musliu (International consultant)

Komuna e Prishtinës (Municipality of Prishtina) 

• Sokol Sallahu (Consultant for urban traffic)

Golden Taxi

• Rexhep Demolli (Consultant for traffic routes / destinations)

External Collaborators

• Dr. Techn. Arben Ahmeti (Researcher)

Publications 

• Atlantik Limani, Kadri Sylejmani, Uran Lajçi, Elvir Misini, Erzen Krasniqi and Bujar Krasniqi, "Smart Traffic Signal Optimization with an Artificial Bee Colony Algorithm", 16-th International Conference on Agents and Artificial Intelligence, Rome, Italy | 24-26 February 2024 (https://icaart.scitevents.org/)

• Elvir Misini, Uran Lajçi, Kadri Sylejmani, Atlantik Limani, Lavdim Kurtaj, Arben Ahmeti, Erzen Krasniqi, “Solving the Traffic Signaling Problem Using the Iterated Local Search Metaheuristic”, Operations Research Forum (under review)

• Uran Lajçi, Atlantik Limani, Elvir Misini, Fjolla Gashi, Kadri Sylejmani, and Arben Ahmeti, “Metaheuristic techniques for automated traffic light scheduling to minimize commute time”, 14th International Conference on the Practice and Theory of Automated Timetabling, Copenhagen, Denmark (to be presented in August 2024)

Developed software

• Artificial Bee Colony based solver for traffic light scheduling (City of Prishtina)

• Artificial Bee Colony based solver for traffic light scheduling (Google Hash Code Variant)

• Iterated Local Search based solver for traffic light scheduling (Google Hash Code Variant)

• A proof-of-concept prototype for traffic light scheduling in the city of Prishtina (https://prishtinagreenwave.com/) 

• A Web-based Test Instance Generator for Traffic Light Scheduling

• A web tool for validation of solutions

• A web tool for the graphical presentation of test cases for the traffic light scheduling problem

• A web tool for visualizing urban mobility traffic data for the City of Pristina

• A web tool for visualizing taxi paths

• A Power BI portal for visualizing urban mobility data of the City of Prishtina

• Solution visualizer for Traffic Signalling Problem

• A web tool for tabular visualization of the green wave

Test instances

• An extended test comprising 48 instances for the Traffic Signaling Problem as defined in Google Hash Code 2021 (qualification round), with 43 new test instances and 5 instances defined in the competition.

• A test set of 4 real-life instances for the City of Prishtina

Solution results

• Solutions from the Iterated Local Search solver for the Traffic Signaling Problem as defined in Google Hash Code 2021 (qualification round)