arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.20188cs.ROcs.AI
收藏

异步快速-慢速视觉-语言-动作策略用于全身机器人操作

Asynchronous Fast-Slow Vision-Language-Action Policies for Whole-Body Robotic Manipulation

AstriBot Team(AstriBot团队)

Teqiang Zou, Hongliang Zeng, Yuxuan Nong, Yifan Li, Kehui Liu, Haotian Yang, Xinyang Ling, Xin Li, Lianyang Ma

AI总结:

DuoCore-FS通过异步快速-慢速框架提升全身机器人操作的实时性能与任务成功率。

摘要:

大多数视觉-语言-动作(VLA)系统将视觉-语言模型(VLM)用于语义推理与生成连续动作信号的动作专家相结合,但两者通常以单一统一频率运行。因此,策略性能受到大VLM推理速度的限制。这种强制同步执行严重限制了全身机器人操作中的控制稳定性和实时性能,该操作涉及更多的关节、更大的运动空间和动态变化的视角。我们介绍了一个真正的异步快速-慢速VLA框架(DuoCore-FS),将系统分为高速路径用于高频动作生成和慢速路径用于丰富的VLM推理。该系统有两个关键特征。首先,一个潜在表示缓冲区连接慢速和快速系统。它存储与场景指令上下文对齐的指令语义和动作推理表示,为快速路径提供高层指导。其次,一个全身动作标记器提供紧凑、统一的全身动作表示。重要的是,VLM和动作专家仍然联合端到端训练,保持统一的策略学习,同时启用异步执行。DuoCore-FS支持3B参数的VLM,同时实现30Hz的全身动作块生成,大约是具有类似模型规模的先前VLA模型的三倍快。现实中的全身操作实验显示,与同步快速-慢速VLA基线相比,任务成功率有所提高,并且响应性显著增强。DuoCore-FS的实现,包括训练、推理和部署,由Astripot作为Astripot机器人平台的一部分提供给商业用户。

英文摘要:

Most Vision-Language-Action (VLA) systems integrate a Vision-Language Model (VLM) for semantic reasoning with an action expert generating continuous action signals, yet both typically run at a single unified frequency. As a result, policy performance is constrained by the low inference speed of large VLMs. This mandatory synchronous execution severely limits control stability and real-time performance in whole-body robotic manipulation, which involves more joints, larger motion spaces, and dynamically changing views. We introduce a truly asynchronous Fast-Slow VLA framework (DuoCore-FS), organizing the system into a fast pathway for high-frequency action generation and a slow pathway for rich VLM reasoning. The system is characterized by two key features. First, a latent representation buffer bridges the slow and fast systems. It stores instruction semantics and action-reasoning representation aligned with the scene-instruction context, providing high-level guidance to the fast pathway. Second, a whole-body action tokenizer provides a compact, unified representation of whole-body actions. Importantly, the VLM and action expert are still jointly trained end-to-end, preserving unified policy learning while enabling asynchronous execution. DuoCore-FS supports a 3B-parameter VLM while achieving 30 Hz whole-body action-chunk generation, approximately three times as fast as prior VLA models with comparable model sizes. Real-world whole-body manipulation experiments demonstrate improved task success rates and significantly enhanced responsiveness compared to synchronous Fast-Slow VLA baselines. The implementation of DuoCore-FS, including training, inference, and deployment, is provided to commercial users by Astribot as part of the Astribot robotic platform.

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。

↑