Foresee: Attentive Future Projections of Chaotic Road Environments

Anil Sharma, Arun Balaji Buduru

Abstract: In this paper, we train a recurrent neural network to learn dynamics of a chaotic road environment and to project the future of the environment on an image. Future projection can be used to anticipate an unseen environment for autonomous driving. Road environment is highly dynamic and complex due to the interaction among traffic participants such as vehicles and pedestrians. Even in such a complex environment, a human driver can easily anticipate the environment and is efficacious to drive safely on the chaotic roads. Proliferation in deep learning research has shown the efficacy of neural networks in learning this kind of human behavior. In the same direction, we investigate recurrent neural networks to understand the road environment. We propose Foresee, a unidirectional gated recurrent units (GRUs) network with attention to project future of the environment in the form of images. We have collected several videos on Delhi roads consisting of various traffic participants, background and infrastructure differences (like 3D pedestrian crossing) at various times on various days. We show that our proposed model performs better than state of the art methods (prednet[9], Enc. Dec. LSTM[15]).