Adaptive Urban Pathfinding Using Deep Learning and Heuristic Optimization

City routing in fast-developing urban cities, such as Cairo, is confronted with serious challenges attributed to dynamic traffic flows, road closures, and variations in weather conditions. This study proposes an inclusive solution for urban pathfinding by incorporating state-of-the-art heuristic algorithms, such as Tabu Search and Branch and Cut, and deep learning models. The techniques utilize real-time traffic and weather data to dynamically optimize routes. We propose an enhanced A* algorithm that adapts its heuristic function based on current environmental conditions and a neural network model that forecasts the next optimal path segment given past and real-time data. An indepth evaluation was made applying the SUMO traffic simulation to evaluate the performance of numerous algorithms and deep learning models, including Multi-Layer Perceptron (MLP), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), Autoencoders, Transformers, Tabu Search, and Branch and Cut. The models were validated in a SUMO-based simulated urban road environment model of Cairo’s road network, where the neural network model cut travel times by 35%, the enhanced A* algorithm by 30%, Tabu Search by 28%, and Branch and Cut by 25% compared to traditional methods. The results demonstrate the promise of integrating sophisticated algorithms with deep learning techniques to improve real-time city navigation, which helps provide more elastic and efficient r outing f or Egyptian cities.