Deep Reinforcement Learning for Optimizing Route Planning in Urban Traffic

Urban traffic networks are efficient route design problems due to the intricacy and dynamic character of contemporary cities. Conventional methods hamper real time adjustment to continually changing traffic conditions, thereby providing suboptimal routes, a greater burden on the roadways, and greater air pollution. This thesis presents a novel deep reinforcement learning (DRL) method to optimize the route planning for urban traffic networks. With this rendering of real time traffic data, we use DRL to make autonomous decisions to reduce travel time and congestion of routes. One advantage of using sophisticated neural networks for feature extraction and policy learning is that, since the model is so adaptable to changing traffic patterns, the model guarantees that. And, in comparison to common routing algorithms, the results of simulations using real world traffic data show that this method is able to greatly enhance route efficiency and overall traffic flow management. The suggested DRL based system is feasible as a smart city program based on intelligent transportation system which can scale to grand metropolitan regions. The research presents that DRL can enhance the commuter run experience, reduce the environmental effect, and alter urban transportation. Future study will also investigate how this system can be better optimized when integrated with traffic infrastructure enabled by the Internet of Things.