Search papers, labs, and topics across Lattice.
This paper introduces OWMDrive, an end-to-end autonomous driving framework that utilizes a 4D Occupancy World Model to forecast multi-step 3D occupancy and enhance decision-making in complex traffic environments. By explicitly modeling temporal causal dynamics and future scene evolution, OWMDrive enables more robust trajectory generation, addressing the limitations of existing static representation methods. Experimental results show that OWMDrive significantly improves planning reliability and safety in high-uncertainty scenarios, outperforming traditional approaches.
OWMDrive's innovative use of a 4D Occupancy World Model allows for foresighted trajectory planning that adapts to dynamic traffic conditions, enhancing safety and reliability.
Autonomous driving systems are steadily moving toward end-to-end paradigms to mitigate the limited adaptability of rule-based pipelines in complex traffic environments. However, most existing learning-based methods still make decisions from static representations of the current scene, without explicit future rollouts or modeling of the temporal causal dynamics in traffic interactions. This limitation often results in unstable or overly conservative planning under high-uncertainty conditions, such as occlusions and unexpected events. To overcome these challenges, we introduce OWMDrive, a generative end-to-end driving framework built upon an Occupancy World Model for multi-step 3D occupancy forecasting, which serves as a conditional prior to guide diffusion-based planning. Conditioned on both current observations and predicted future states, the planner iteratively refines trajectory candidates to generate a reinforced driving trajectory. By explicitly modeling scene evolution over future horizons, OWMDrive captures key spatiotemporal causal dependencies, which leads to more foresighted and robust trajectory generation. Extensive experiments demonstrate that OWMDrive significantly improves planning reliability and safety, especially in challenging and partially observable driving scenarios.
