MotionCNN: A Strong Baseline for Motion Prediction in Autonomous Driving

TL;DR

MotionCNN introduces a simple convolutional neural network-based approach for multimodal motion prediction in autonomous driving, achieving competitive results and ranking third in a major challenge.

Contribution

It presents a straightforward, easy-to-implement CNN-based baseline for motion prediction that rivals more complex state-of-the-art methods.

Findings

Achieves competitive performance on the Waymo Open Dataset

Ranks 3rd in the 2021 Waymo Motion Prediction Challenge

Source code is publicly available for reproducibility

Abstract

To plan a safe and efficient route, an autonomous vehicle should anticipate future motions of other agents around it. Motion prediction is an extremely challenging task that recently gained significant attention within the research community. In this work, we present a simple and yet very strong baseline for multimodal motion prediction based purely on Convolutional Neural Networks. While being easy-to-implement, the proposed approach achieves competitive performance compared to the state-of-the-art methods and ranks 3rd on the 2021 Waymo Open Dataset Motion Prediction Challenge. Our source code is publicly available at GitHub