A Survey of Motion Prediction for Autonomous Vehicles Using the Lyft Dataset

Recent years have seen Autonomous Vehicles (AVs) develop dramatically. With industry and academia pushing for research, perception, planning, and control in AVs have evolved rapidly. There are still significant challenges to be overcome before AVs can navigate busy roads without human assistance. One of these challenges is the motion estimation of other traffic agents (cars, pedestrians, and cyclists), and analysing such behaviour requires a comprehensive understanding of numerous variables. In this paper, we study how the Lyft Dataset contributed to predicting the trajectories of various traffic agents using different computer vision models as backbones. We also attempt to implement a similar model and compare it with existing models.