Drive-JEPA: Video JEPA Meets Multimodal Trajectory Distillation for End-to-End Driving

TL;DR

Drive-JEPA introduces a novel framework combining video pretraining and multimodal trajectory distillation to significantly improve end-to-end autonomous driving performance, achieving state-of-the-art results on benchmark datasets.

Contribution

It adapts V-JEPA for driving video pretraining and proposes a proposal-centric planner with trajectory distillation for better multimodal behavior learning.

Findings

Outperforms prior methods by 3 PDMS in perception-free setting

Achieves 93.3 PDMS on v1 and 87.8 EPDMS on v2 datasets

Sets new state-of-the-art in end-to-end driving benchmarks

Abstract

End-to-end autonomous driving increasingly leverages self-supervised video pretraining to learn transferable planning representations. However, pretraining video world models for scene understanding has so far brought only limited improvements. This limitation is compounded by the inherent ambiguity of driving: each scene typically provides only a single human trajectory, making it difficult to learn multimodal behaviors. In this work, we propose Drive-JEPA, a framework that integrates Video Joint-Embedding Predictive Architecture (V-JEPA) with multimodal trajectory distillation for end-to-end driving. First, we adapt V-JEPA for end-to-end driving, pretraining a ViT encoder on large-scale driving videos to produce predictive representations aligned with trajectory planning. Second, we introduce a proposal-centric planner that distills diverse simulator-generated trajectories alongside human trajectories, with a momentum-aware selection mechanism to promote stable and safe behavior. When evaluated on NAVSIM, the V-JEPA representation combined with a simple transformer-based decoder outperforms prior methods by 3 PDMS in the perception-free setting. The complete Drive-JEPA framework achieves 93.3 PDMS on v1 and 87.8 EPDMS on v2, setting a new state-of-the-art.