Waymo Open Motion Dataset

The Waymo Open Motion Dataset (WOMD) is a large-scale, publicly available dataset of real-world driving scenarios used to advance research in autonomous driving, particularly in motion prediction and scene understanding. Current research focuses on developing sophisticated models, including transformer-based architectures and diffusion models, to accurately predict multi-modal future trajectories of multiple agents, often incorporating multi-sensor data fusion (camera and LiDAR) and addressing challenges like handling interactions and mitigating prediction errors. WOMD's comprehensive nature and associated challenges are driving significant progress in the field, leading to improved algorithms for autonomous vehicle perception, planning, and simulation.