Realtime-VLA V2:学习以快速、平滑和准确地运行VLAs
Realtime-VLA V2: Learning to Run VLAs Fast, Smooth, and Accurate
AI总结:
本文提出Realtime-VLA V2,通过校准、规划与控制及学习方法,实现机器人在真实任务中高速、高精度和高灵活性的VLAs运行。
摘要:
在将VLAs模型部署到真实机器人任务中时,执行速度至关重要。在之前的arXiv:2510.26742工作中,我们分析了如何在GPU上使VLAs的神经计算快速运行。然而,我们留有如何实际在真实机器人上部署VLAs系统的问题。在本报告中,我们描述了一套实用技术,以实现端到端的结果:在需要准确性和灵活性的真实任务中,运行由VLAs驱动的机器人以令人印象深刻的速度。技术栈涵盖校准、规划与控制,以及基于学习的方法来识别最优执行速度。在所展示的任务中,机器人甚至以与随意人类操作相当的速度运行,并接近我们轻量臂的硬件极限。未加速的视频和推理跟踪可在https://dexmal.github.io/realtime-vla-v2/上获得。
英文摘要:
In deployment of the VLA models to real-world robotic tasks, execution speed matters. In previous work arXiv:2510.26742 we analyze how to make neural computation of VLAs on GPU fast. However, we leave the question of how to actually deploy the VLA system on the real robots open. In this report we describe a set of practical techniques to achieve the end-to-end result of running a VLA-driven robot at an impressive speed in real world tasks that require both accuracy and dexterity. The stack of technology ranges across calibration, planning & control, and learning based method to identify optimal execution speed. In the tasks we show, the robot even executes in a speed on par with casual human operation and approaching the hardware limit of our lightweight arm. The unaccelerated videos and inference traces are provided in https://dexmal.github.io/realtime-vla-v2/.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。