Vine-inspired robots achieve large workspace coverage through tip eversion, enabling safe navigation in confined and cluttered environments. However, their deployment in free space is fundamentally limited by low axial stiffness, poor load-bearing capacity, and the inability to retain shape during and after steering. In this work, we propose a reconfigurable pneumatic joint (RPJ) architecture that introduces discrete, pressure-tunable stiffness along the robot body without compromising continuous growth. Each RPJ module comprises symmetrically distributed pneumatic chambers that locally increase bending stiffness when pressurized, enabling decoupling between global compliance and localized rigidity. We integrate the RPJs into a soft growing robot with tendon-driven steering and develop a compact base station for mid-air eversion. System characterization and experimental validation demonstrate moderate pressure requirements for eversion, as well as comparable localized stiffening and steering performance to layer-jamming mechanisms. Demonstrations further show that the proposed robot achieves improved shape retention during bending, reduced gravitational deflection under load, cascading retraction, and reliable payload transport up to 202 g in free space. The RPJ mechanism establishes a practical pathway toward structurally adaptive vine robots for manipulation-oriented tasks such as object sorting and adaptive exploration in unconstrained environments.

翻译：摘要：基于藤蔓仿生的机器人通过尖端外翻实现大范围工作空间覆盖，能够在狭小杂乱环境中安全导航。然而，其在自由空间中的部署受到低轴向刚度、较差承载能力以及在转向过程中及转向后无法保持形状的根本性限制。本文提出一种可重构气动关节架构，可在不牺牲连续生长能力的前提下，沿机器人本体引入离散的、气压可调的刚度。每个RPJ模块包含对称分布的气动腔室，在加压时局部增加弯曲刚度，从而实现全局柔顺性与局部刚性的解耦。我们将RPJ集成到带有腱绳驱动的软体生长机器人中，并开发了紧凑型基站以实现空中外翻。系统特性表征与实验验证表明，外翻所需气压适中，且局部刚化和转向性能与层夹机制相当。进一步的演示表明，所提出的机器人在弯曲过程中实现了增强的形状保持能力、负载下重力引起的偏转减小、级联回缩以及在自由空间中可靠运输高达202克的有效载荷。RPJ机制为面向操控任务的结构自适应藤蔓机器人（如在无约束环境中进行物体分类与自适应探索）提供了一条实用路径。