Realistic World Model for Autonomous Driving: Integrating Physical Constraints and Multi-agent Interactions
Keywords: Autonomous Driving, World Model, Trajectory Forecasting, Motion Planning
Abstract: Ensuring safety in autonomous driving, particularly in complex and dynamic environments, remains a significant challenge. To address this issue, we propose a novel traffic world model. While existing trajectory forecasting methods typically focus on predicting individual agents and may neglect critical factors such as vehicle dimensions, orientation, and physical constraints, our model incorporates these elements comprehensively. Unlike previous methods that often result in unrealistic scenarios such as collisions or off-road driving, our model integrates physical constraints and introduces innovative loss functions—including safe distance loss and road departure loss—to ensure that the generated trajectories are both realistic and feasible. By simultaneously predicting the trajectories of all agents and explicitly modeling interactions across various scenarios, our approach significantly enhances realism and safety. Our world model functions as a generator, simulator, and trajectory forecasting tool, demonstrating substantial improvements over traditional methods and achieving competitive performance in reducing collision and off-road rates.
Primary Area: learning on time series and dynamical systems
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3475
Loading