PGX 框架提升 VLA 人形机器人长时程移动操作成功率
PGX: Physically Grounded Execution for Long-Horizon Humanoid Loco-Manipulation
arXiv 上的这篇机器人论文提出了 PGX,用状态机加世界坐标锁定两个免训练组件,把 Unitree G1 上的长时程操作任务拆开做,VLA 只管技能执行。做机器人部署的值得看看这个分层思路。
PGX(Physically Grounded Execution)是一个针对 VLA 人形机器人长时程移动操作(loco-manipulation)的框架,核心思路是把时序控制与精度控制从单体 VLA 模型中剥离出来,交给两个免训练的经典组件处理,让 VLA 专注于各阶段的具体技能动作。
框架分三层:第一层用基于谓词(predicate)的状态机管理任务进程,以物理谓词门控阶段切换、触发恢复并协调导航与等待;第二层在主动操作阶段由平坦或按阶段训练的 VLA 模型执行任务相关动作;第三层通过世界坐标锁定(WCL),借助基站保持、物体相对定位和世界锚定接触目标来维持空间一致性。同一套谓词语义贯穿离线演示分割、在线执行和分阶段评估。
实验覆盖三个仿真任务和一个真实任务,使用 Unitree G1 人形机器人,在阶段推进和任务成功率上均有提升。对做机器人 VLA 部署的开发者来说,这种分层架构提供了一条不改模型、不微调就能拉长任务时程的工程路线。
PGX: Physically Grounded Execution for Long-Horizon Humanoid Loco-Manipulation
Robotic vision-language-action (VLA) models can execute diverse individual manipulation skills, yet they still struggle with long-horizon loco-manipulation tasks, where temporal structure, skills, and precision are tightly coupled within a monolithic model. We propose Physically Grounded Execution (PGX), a framework that offloads temporal and precision control to two classical training-free components, allowing the VLA to focus on stage-specific skill dynamics. PGX decomposes long-horizon loco-manipulation into three layers: (i) a predicate-based state machine governs task progression, using physical predicates to gate phase transitions, trigger recovery, and coordinate navigation and waiting. Shared predicate semantics connect offline demonstration segmentation, online execution, and stage-wise evaluation; (ii) VLA skills operate within active manipulation phases, with flat or stage-trained models concentrating on task-relevant actions; and (iii) world-coordinate locking (WCL) maintains spatial consistency through base station keeping, object-relative positioning, and world-anchored contact targets. We demonstrate the effectiveness of PGX with extensive experiments conducted on three simulated tasks and one real-world task using a Unitree G1 humanoid robot for long-horizon loco-manipulation, showing improvements in stage progression and task success rates. Project page is available at: https://kl-research.github.io/pgx
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include PGX, VLA, Unitree G1, humanoid robots.
Source attribution: curated from arXiv cs.AI (@Kai Lu, Chenyang Ma, Ioannis Havoutis).