外观
论文arXiv 2609.062511 个月前
MobileVLA-R1 2.0：用于移动机器人控制的RL增强推理
MobileVLA-R1 2.0用监督CoT对齐与强化学习耦合具身推理和控制，并在导航、四足控制和人形移动操作上验证。
宇树 Go2原文链接：MobileVLA-R1 2.0: RL-Enhanced Reasoning for Mobile Robot Control（arXiv 2609.06251 · 发布日期：2026-09-05）
MobileVLA-R1 2.0: RL-Enhanced Reasoning for Mobile Robot Control
MobileVLA-R1 2.0用监督CoT对齐与强化学习耦合具身推理和控制，并在导航、四足控制和人形移动操作上验证。
核心要点
- 自然语言指令到可靠可执行动作仍受高层语义推理与底层控制鸿沟制约。
- 框架通过监督CoT对齐和强化学习，学习具身轨迹上的多粒度推理。
- 推理条件动作解码器将多模态推理映射为任务级目标，再由控制器转为具体命令。
- 在VLN-CE、QUARD和真实部署中优于强VLA基线，并提升SR与G1全任务成功率。
验证范围与局限：验证范围包括语言引导导航、四足控制，以及Unitree Go2和G1机器人的真实世界部署。
英文摘要 (Abstract)
Grounding natural-language instructions into reliable and executable actions remains a fundamental challenge for vision-language-action (VLA) systems on mobile robots, due to the persistent gap between high-level semantic reasoning and low-level locomotion and manipulation control. Existing approaches often rely on implicit reasoning or monolithic action prediction, making it difficult to maintain coherent long-horizon decision making while producing precise and adaptable robot actions. To address this challenge, we propose MobileVLA-R1 2.0, an RL-enhanced VLA framework that explicitly couples structured embodied reasoning with executable mobile robot control. The framework learns multi-granularity reasoning over embodied trajectories through supervised Chain-of-Thought (CoT) alignment and reinforcement learning, improving reasoning-to-action consistency beyond purely behavioral supervision. To support both locomotion and manipulation, we further introduce a reasoning-conditioned action decoder that maps multimodal reasoning representations to task-level action targets, which are subsequently translated into embodiment-specific commands by robot controllers. This design provides a unified perception-reasoning-action interface while decoupling high-level action generation from robot-specific actuation. We conduct extensive evaluations on language-guided navigation, quadruped control, and humanoid mobile manipulation, covering VLN-CE, QUARD, and real-world deployments on Unitree Go2 and G1 robots. MobileVLA-R1 2.0 consistently outperforms strong VLA baselines, achieving an average 1.6 point improvement in SR on VLN-CE and a 10.0 point improvement in full-task success on real-world G1 mobile manipulation tasks over MobileVLA-R1, while demonstrating robust long-horizon instruction following and closed-loop execution across different robotic platforms.
作者：Ting Huang, Yue Huang, Zeyu Zhang, Shuicheng Yan, Hao Tang
来源
本页事实对应的公开页面，以原文为准。