Dynamic Trajectory Planning for Urban Traffic Using Deep Learning-Based Traffic Prediction

Traditional trajectory planning systems in navigation applications generate routes based on the current traffi conditions at the time of departure. However, traffi dynamics are highly variable, and aroute that is initially optimal may become congested during the trip, leading to suboptimal travel timesand traffi jams. This paper proposes a novel approach to dynamic trajectory planning by integratingdeep learning-based traffi prediction models into the routing process. We leverage neural networksdesigned for time series forecasting to predict short-term traffi patterns, incorporating factors suchas vehicle velocity and local traffi densities. By forecasting the future state of candidate routes, ourmethod continuously updates the optimal trajectory in real time, ensuring improved adaptability tochanging traffi conditions. We evaluate the performance of different time series models, including re-current neural networks, long short-term memory networks, and temporal convolutional networks, using real-world urban traffi datasets. Experimental results demonstrate that incorporating traffi predictionsignificantl enhances route optimality and reduces overall travel time compared to static trajectoryplanning approaches.