Based on interlimb neural coupling, gait robotic systems should produce walking-like movement in both upper and lower limbs for effective walking restoration. Two orthoses were previously designed in our lab to provide passive walking with arm swing. However, an active system for walking with arm swing is desirable to serve as a testbed for investigation of interlimb neural coupling in response to voluntary input. Given the important function of the ankle joint during normal walking, this work aimed to develop an improved rotational orthosis for walking with arm swing, which is called ROWAS II, and especially to develop and evaluate the algorithms for active ankle control. After description of the mechanical structure and control schemes of the overall ROWAS II system, the closed-loop position control and adjustable admittance control algorithms were firstly deduced, then simulated in Matlab/Simulink and finally implemented in the ROWAS II system. Six able-bodied participants were recruited to use the ROWAS II system in passive mode, and then to estimate the active ankle mechanism. It was showed that the closed-loop position control algorithms enabled the ROWAS II system to track the target arm-leg walking movement patterns well in passive mode, with the tracking error of each joint <0.7°. The adjustable admittance control algorithms enabled the participants to voluntarily adjust the ankle movement by exerting various active force. Higher admittance gains enabled the participants to more easily adjust the movement trajectory of the ankle mechanism. The ROWAS II system is technically feasible to produce walking-like movement in the bilateral upper and lower limbs in passive mode, and the ankle mechanism has technical potential to provide various active ankle training during gait rehabilitation. This novel ROWAS II system can serve as a testbed for further investigation of interlimb neural coupling in response to voluntary ankle movement and is technically feasible to provide a new training paradigm of walking with arm swing and active ankle control.

基于肢体神经耦合，步态机器人系统应在上下肢中产生类​​似步行的运动，以有效地恢复步行。我们实验室中先前设计了两个矫形器，以提供手臂摆动的被动行走。然而，期望有一种用于主动摆动的主动系统，以作为试验床，用于研究对自主输入做出响应的肢体神经耦合。考虑到脚踝关节在正常行走过程中的重要功能，这项工作旨在开发一种改进的旋转矫正器，用于伴随手臂摆动的行走，称为ROWAS II，尤其是开发和评估主动脚踝控制算法。在描述整个ROWAS II系统的机械结构和控制方案之后，首先推导了闭环位置控制和可调导纳控制算法，然后在Matlab / Simulink中进行仿真，最后在ROWAS II系统中实现。招募了六名身体强健的参与者，以被动模式使用ROWAS II系统，然后评估主动踝关节机制。结果表明，闭环位置控制算法使ROWAS II系统能够在被动模式下很好地跟踪目标臂-腿步行运动模式，每个关节的跟踪误差<0.7°。可调的导纳控制算法使参与者能够通过施加各种作用力来自动调整脚踝的运动。较高的导纳增益使参与者能够更轻松地调整脚踝机构的运动轨迹。ROWAS II系统在技术上是可行的，以被动方式在双侧上肢和下肢产生类似步行的运动，并且脚踝机制具有在步态康复期间提供各种主动式脚踝训练的技术潜力。这种新颖的ROWAS II系统可以用作进一步研究肢体间神经耦合以响应自发性踝关节运动的试验平台，并且在技术上可行以提供一种新的手臂摆动和主动踝关节控制的训练范例。