# Vision-Language Cross-Attention for Real-Time Autonomous Driving

arXiv: https://arxiv.org/abs/2507.23064 · HTML: https://mlsanity.com/paper/2507.23064/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in VLMs for autonomous driving, one of 9 directions in End-to-end autonomous driving.
- Builds on 22 of its 25 references that are in the map.
- First posted on arXiv 30 Jul 2025.

## Placed in

- End-to-end autonomous driving / [VLMs for autonomous driving](https://mlsanity.com/driving/vlms-for-autonomous-driving/)

## Builds on (most cited first, 22 of 22)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- ["Why Should I Trust You?": Explaining the Predictions of Any Classifier](https://mlsanity.com/paper/1602.04938/) (2016)
- [nuScenes: A multimodal dataset for autonomous driving](https://mlsanity.com/paper/1903.11027/) (2019)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [Scalability in Perception for Autonomous Driving: Waymo Open Dataset](https://mlsanity.com/paper/1912.04838/) (2019)
- [Do As I Can, Not As I Say: Grounding Language in Robotic Affordances](https://mlsanity.com/paper/2204.01691/) (2022)
- [PaLM-E: An Embodied Multimodal Language Model](https://mlsanity.com/paper/2303.03378/) (2023)
- [Sanity Checks for Saliency Maps](https://mlsanity.com/paper/1810.03292/) (2018)
- [Vision-and-Language Navigation: Interpreting visually-grounded navigation instructions in real environments](https://mlsanity.com/paper/1711.07280/) (2017)
- [ChauffeurNet: Learning to Drive by Imitating the Best and Synthesizing the Worst](https://mlsanity.com/paper/1812.03079/) (2018)
- [MapTR: Structured Modeling and Learning for Online Vectorized HD Map Construction](https://mlsanity.com/paper/2208.14437/) (2022)
- [HDMapNet: An Online HD Map Construction and Evaluation Framework](https://mlsanity.com/paper/2107.06307/) (2021)
- [VectorMapNet: End-to-end Vectorized HD Map Learning](https://mlsanity.com/paper/2206.08920/) (2022)
- [Textual Explanations for Self-Driving Vehicles](https://mlsanity.com/paper/1807.11546/) (2018)
- [Computer Vision for Autonomous Vehicles: Problems, Datasets and State of the Art](https://mlsanity.com/paper/1704.05519/) (2017)
- [Explainable Artificial Intelligence for Autonomous Driving: A Comprehensive Overview and Field Guide for Future Research Directions](https://mlsanity.com/paper/2112.11561/) (2021)
- [DURA-CPS: A Multi-Role Orchestrator for Dependability Assurance in LLM-Enabled Cyber-Physical Systems](https://mlsanity.com/paper/2506.06381/) (2025)
- [PhysNav-DG: A Novel Adaptive Framework for Robust VLM-Sensor Fusion in Navigation Applications](https://mlsanity.com/paper/2505.01881/) (2025)

## Abstract (arXiv)

Autonomous cars need geometric accuracy and semantic understanding to navigate complex environments, yet most stacks handle them separately. We present XYZ-Drive, a single vision-language model that reads a front-camera frame, a 25m × 25m overhead map, and the next waypoint, then outputs steering and speed. A lightweight goal-centered cross-attention layer lets waypoint tokens highlight relevant image and map patches, supporting both action and textual explanations, before the fused tokens enter a partially fine-tuned LLaMA-3.2 11B model. On the MD-NEX Outdoor-Driving benchmark XYZ-Drive attains 95% success and 0.80 Success weighted by Path Length (SPL), surpassing PhysNav-DG by 15%. and halving collisions, all while significantly improving efficiency by using only a single branch. Sixteen ablations explain the gains. Removing any modality (vision, waypoint, map) drops success by up to 11%, confirming their complementary roles and rich connections. Replacing goal-centered attention with simple concatenation cuts 3% in performance, showing query-based fusion injects map knowledge more effectively. Keeping the transformer frozen loses 5%, showing the importance of fine-tuning when applying VLMs for specific tasks such as autonomous driving. Coarsening map resolution from 10 cm to 40 cm blurs lane edges and raises crash rate. Overall, these results demonstrate that early, token-level fusion of intent and map layout enables accurate, transparent, real-time driving.