Traffic Agent Movement Prediction Using ResNet-based Model

Autonomous driving is a promising field, which brings conveniences to the life of people and optimizes the operations of the social system. Although many advantages it has, the complexity of autonomous driving hinders the applications of it in practice. autonomous driving is a comprehensive and complex project, which contains lots of difficult challenges. And the traffic agent movement prediction is one of them. In this paper, we regard the traffic agent movement prediction as a regression problem. And a deep neural network model of which the backbone is ResNet101 is proposed to deal with the regression. To demonstrate the efficiency of the proposed method, experiments on Lyft Motion Prediction for Autonomous Vehicles data set are conducted. And the quantitative comparisons of the experimental results indicate that the proposed method is more efficient on the traffic motion prediction than comparing methods.