With the rapid increase in urban populations, the proliferation of private vehicles, and the worsening state of air quality, conventional urban transportation planning methods are struggling to meet modern demands. These traditional systems often lack the adaptability and intelligence required to respond effectively to dynamic and complex traffic patterns. In response to these challenges, this study proposes a reinforcement learning (RL)-based approach designed to enhance urban transportation efficiency and sustainability. The core objective of the proposed model is to determine optimal routes between origin and destination points by dynamically avoiding congested areas. Unlike static routing algorithms, the RL model continuously learns and adapts to traffic conditions, enabling the selection of routes that minimize travel time and reduce vehicle idling. As a result, the approach significantly contributes to lowering fossil fuel consumption and energy use, while simultaneously addressing the broader environmental concern of urban air pollution. The integration of artificial intelligence in transportation systems through RL not only enhances service quality and traffic flow but also supports the development of smarter, greener cities. This study underscores the transformative potential of RL in revolutionizing traffic management systems and presents a viable framework for future intelligent transportation networks.