With the developments in autonomous driving and the popularity, the subject has gained over the past few years in the scientific communities, especially considering that most of the traffic accidents are due to recognition errors and perception being neglected in the literature, this study aimed to develop a prediction method and employed it in intersection driving scenario. Being a data-driven approach, a simulation was set up in SUMO, and a combination of Dynamic Bayesian Network and

more »

... Neural Network were tasked to make predictions for the states of the ego vehicle and that of other traffic participants for 1, 2, 5, and 10-second horizons. A Kalman filter was used as a post-processing measure to ensure smooth transitions in the velocities especially in longer horizons. Results, disclosed for a random traffic participant pair in the data pool, exhibit valid predictions and competent accuracy.