Sequential Rasterized Image-based Trajectory Prediction Deep-Learning Model

In this paper, we design an ambient vehicle path prediction model based on deep learning. The most important goal of the autonomous driving system is to ensure the safety of passengers. Therefore, it is essential to predict changes in the surrounding environment of vehicles. We generate raster images to take into account road conditions and vehicles, which are moving objects in driving environments. And we use a pair of sequential images rather than a single image as input to the deep learning model. In addition, speed, acceleration, and change of heading rate are used together as input to a deep learning model to provide status information on the vehicle of interest to infer routes. Through this study, it was confirmed that providing sequential information on the road environment contributes to improving the performance of the trajectory prediction by using sequential images as input data for the deep learning model.