THESIS:Sustainable Urban Mobility:Challenges and Technological Approaches

This thesis presents a comprehensive study of the landscape of Sustainable Urban Mobility. Sustainable Urban Mobility represents a modern approach to organizing urban transportation, aiming to develop a transport system that is environmentally friendly, efficient, and human-centric. It focuses on modes of transport that contribute to reducing pollution, noise levels, and carbon dioxide (CO2) emissions, while enhancing the health and safety of residents and prioritizing people over vehicles. To implement the framework of Sustainable Urban Mobility across a wide range of cities and countries, specific changes and appropriate measures are required. This thesis examines the subject in depth, beginning with an analysis of the United Nations Agenda for Sustainable Development and the seventeen Sustainable Development Goals (SDGs) it defines. Subsequently, the concept of Sustainable Urban Mobility is thoroughly explained, focusing primarily on an extensive analysis of the needs and challenges facing sustainable mobility in modern societies. Emphasis is then placed on the presentation, explanation, and analysis of various proposals and policies aimed at the further establishment and promotion of Sustainable Urban Mobility. In this context, a categorization and modeling of various types of cities and urban areas in modern societies is performed. Through a statistical analysis of mobility indicators for each category, the thesis ensures that the proposed mobility measures and recommendations are representative, entirely appropriate, and accessible for each specific category

1.Urban Classification of Cities

This project provides a computational decision-making tool designed to bridge the gap between theoretical urban categorization and practical Sustainable Urban Mobility Plans (SUMPs). By employing Data Engineering and Pattern Recognition principles, urban environments are modeled using Vector Space Classification. Each city is represented as a normalized 7-dimensional vector (incorporating metrics like population, GDP, traffic congestion, and innovation). The algorithm uses Euclidean Distance to classify any given city against 8 mathematically defined urban archetypes (Cluster Centroids) inspired by the Brookings Institution.

2.Smart Traffic Light System

This project features a custom microscopic traffic simulation developed in Python using the Pygame library, designed to visually benchmark two distinct traffic light control strategies: a traditional static timer-based system and an adaptive "smart" algorithm. Created as part of an undergraduate thesis, the simulation models a standard intersection where vehicles spawn stochastically and navigate the crossing while strictly adhering to traffic signals and built-in collision avoidance protocols.

The execution of the program is divided into two sequential phases to provide a direct, mathematically sound comparison. In the initial Static Control phase, the intersection operates under a conventional model where the traffic lights toggle at strictly fixed time intervals (STATIC_CYCLE), completely blind to the actual traffic conditions. Once the designated timeframe elapses, the simulation seamlessly resets and transitions into the Smart Control phase. During this phase, the traffic lights dynamically adapt to the environment based on real-time data. The intelligent controller utilizes Gap Logic to immediately switch the signals if a green lane becomes empty while cars are waiting at the red light. Furthermore, it employs Queue Balancing to force a signal change if the queue on the red light grows disproportionately larger than the active lane, all while enforcing a Minimum Green Time to ensure road safety and prevent rapid, chaotic light toggling.

Throughout both simulated phases, the system meticulously tracks the total wait times by monitoring the stationary frames of every single vehicle, ultimately calculating the precise average delay per car. Upon completion of the simulation, the visual environment transitions to a final benchmarking dashboard. This screen provides a side-by-side performance comparison between the two methodologies, clearly displaying the total wait times, the average delay per vehicle, and the overall percentage of improvement achieved by the adaptive algorithm.

3.Fleet Routing and Hybrid Genetic Algorithm Optimization This module introduces a highly constrained route optimization engine designed to solve a complex variant of the Capacitated Vehicle Routing Problem with Time Windows (CVRPTW). Developed as a core component of my thesis, this Python script focuses on optimizing a fleet of transit vehicles (e.g., employee shuttles) navigating toward a central depot. The environment enforces strict, real-world operational constraints: each vehicle has a maximum capacity of 5 passengers, every individual has a specific 15-minute pickup window, and, crucially, no passenger can remain in transit for more than 45 minutes (Maximum Ride Time limit). To navigate this massive search space, the project implements a custom Hybrid Genetic Algorithm (HGA).

The evolutionary process maintains a population of 80 potential routing solutions over 300 generations. It heavily penalizes inefficient fleet sizes through a custom fitness function, ensuring the algorithm prioritizes minimizing the total number of vehicles dispatched before reducing the overall distance traveled. To prevent the algorithm from getting stuck in local optima, the HGA is augmented with a Local Search heuristic injected every 20 generations, which iteratively swaps adjacent stops to fine-tune the routes. To demonstrate its effectiveness, the script simultaneously solves the same dataset using a standard Greedy Algorithm (which simply prioritizes the geographically closest unvisited node) and compares the results. The output is a high-resolution, side-by-side visual comparison generated via matplotlib, detailing the exact routes, pickup times, and ride durations for every passenger, clearly highlighting the superior logistical efficiency and compliance of the Genetic Algorithm.