INTP-DM: Intention-Aware Multimodal Vehicle Trajectory Prediction Using Diffusion Model

Yuanbo Yang, Sheng Jin, Xiaoxiang Na, Panagiotis Angeloudis, Simon Hu

Abstract: The ability to predict the future trajectories of surrounding vehicles is crucial for the safe and efficient operation of autonomous vehicles. However, most existing data-driven trajectory prediction models lack an understanding of the corresponding driving intentions, which makes it challenging to produce interpretable results and often leads to inaccuracies in trajectory prediction. In this paper, we introduce INTP-DM, a multimodal trajectory prediction model that incorporates driving intentions into trajectory prediction. Our model achieves intention-aware trajectory prediction through a dedicated intention recognition module and an advanced diffusion-based trajectory prediction module. Paired with a two-stage training scheme and motion constraints, our method can generate controllable, interpretable trajectories that adhere to physical rules. Experimental results on the Waymo open dataset demonstrate that our method can achieve state-of-the-art performance, particularly in challenging long-trajectory prediction scenarios.