Back to AI News
arxiv

FastRoverVLA: Fast Vision-Language-Action Execution for Off-Road Autonomous Driving

FastRoverVLA introduces a training-free interface that decouples visual plan generation from motion-conditioned execution for off-road vision-language-action models. It maintains decoded plans between visual predictions using a speed-curvature action buffer aligned with measured motion, then applies a bounded refresh rule to assign short or long reuse budgets. Over a 3 s horizon, evaluation on ORAD-3D and RELLIS-3D reports ADE/FDE of 0.328/0.596 m and 0.266/0.526 m, reducing FDE by 18.0% and 36.7% at request rates of 6.0% and 4.8%. The runs also reduce Qwen3-VL request-processing cost by factors of 16.6 and 20.8 under serial replay, with source code and evaluation materials publicly available.

AI Brief

It presents a concrete, code-backed method for reducing expensive multimodal inference in off-road VLA driving while reporting accuracy and cost improvements.

arXiv:2611.26242v1 Announce Type: cross Abstract: Off-road vision-language-action (VLA) models generate scene-conditioned plans, but frequent trajectory updates can repeatedly invoke expensive multimodal inference. Reusing a prediction reduces this demand while leaving its commands exposed to changes in vehicle motion. FastRoverVLA addresses this mismatch by decoupling visual plan generation from motion-conditioned execution. Its training-free interface maintains decoded plans between visual predictions. A speed-curvature action buffer is aligned with measured motion at each update, integrated into a local trajectory, and advanced by consuming the current action. The corrected remainder is written back to the buffer, preserving motion corrections across successive updates. A bounded refresh rule normalizes speed and curvature variations within a newly decoded plan and assigns a short or long reuse budget according to the resulting plan-roughness score. The interface requires neither backbone features nor an additional learned driving policy. Over a 3 s horizon, independent real-request evaluation achieves average displacement error (ADE) and final displacement error (FDE) of 0.328/0.596 m on ORAD-3D and 0.266/0.526 m on RELLIS-3D, reducing FDE by 18.0% and 36.7% at request rates of 6.0% and 4.8%, respectively. These runs retain lower errors across four backbones and both datasets and reduce Qwen3-VL request-processing cost by factors of 16.6 and 20.8 under serial replay, while mechanism and motion-only controls identify persistent state correction as the main source of the observed accuracy benefit. The source code and evaluation materials are publicly available at https://github.com/ADEPT-ORG/FastRoverVLA.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!