Saved in:
Bibliographic Details
Main Authors: Patapati, Santosh, Srinivasan, Trisanth, Ambati, Murari
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.23064
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912643851223040
author Patapati, Santosh
Srinivasan, Trisanth
Ambati, Murari
author_facet Patapati, Santosh
Srinivasan, Trisanth
Ambati, Murari
contents 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 $\times$ 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.
format Preprint
id arxiv_https___arxiv_org_abs_2507_23064
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vision-Language Cross-Attention for Real-Time Autonomous Driving
Patapati, Santosh
Srinivasan, Trisanth
Ambati, Murari
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Robotics
I.4.8; I.2.10; I.2.6; C.3.3; I.4.9
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 $\times$ 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.
title Vision-Language Cross-Attention for Real-Time Autonomous Driving
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Robotics
I.4.8; I.2.10; I.2.6; C.3.3; I.4.9
url https://arxiv.org/abs/2507.23064