MOJITO: Modal Joint Learning for Unified End-to-End Autonomous Driving
This work addresses the information bottleneck in cascaded autonomous driving systems, offering a new paradigm for integrating multi-modal sensor data directly into planning.
MOJITO proposes a unified end-to-end autonomous driving framework that replaces the cascaded perception-to-planning pipeline with block-wise modal joint attention, enabling the planner to directly access multi-modal features. It achieves state-of-the-art performance with 88.9 PDMS on NAVSIM v1 and 88.4 EPDMS on NAVSIM v2.
End-to-end autonomous driving systems commonly follow a cascaded two-stage pipeline where a perception stage compresses multi-modal sensor inputs into a compact context and a downstream planner predicts trajectories conditioned on this context. We argue that this one-way perception-to-planning interface forces sensor inputs into a compact representation, losing the fine-grained details critical for planning. Moreover, by constraining the planner to this compressed context, it is difficult to leverage the rich representations offered by modern vision foundation models. To address these issues, we propose MOJITO, a unified sensor-to-action framework for end-to-end autonomous driving built on modal joint learning. MOJITO removes the cascaded interface and instead performs block-wise Modal Joint Attention that simultaneously updates action, image, and LiDAR features, allowing the planner to directly access multi-modal features during action generation. MOJITO achieves 88.9 PDMS on the NAVSIM v1 dataset and 88.4 EPDMS on the more challenging NAVSIM v2 dataset, setting a new state-of-the-art. Extensive experiments further demonstrate strong scalability, instruction following, and diverse trajectory generation. Code and models are available at https://github.com/mumucc01/MOJITO.