The motion prediction for autonomous vehicles task is complex and relevant in the present world for all the existing applications. The task consists in predicting the moviment of all traffic agents and their respective interactions to being used to plan the autonomous vehicle path. This task has been enabled by the resurgence of neural networks, made possible by the advancements in computational power. This seeks to explore the use and compatiblity of deep neural networks in this task, given that the increase in depth has shown a promising way to improve the performance of other computer vision tasks.