Single degree-of-freedom mechanized gait trainers employing linkage mechanisms can imitate the gait trajectory of the human user at a potentially lower cost than current commercial solutions. Our goal was to evaluate recently introduced four-, six- and eight-bar designs from the standpoint of kinematics, that is, the ability to follow a natural gait pattern. Based on a gait database of 113 healthy individuals, we synthesized mechanisms allowing two link length adjustments and compared trajectory and path accuracy of the ankle position across mechanisms. We used cross-validation to obtain data-independent accuracy measures. We used gait trajectories from a large human database and predicted additional trajectories to represent the effects of varying heights and speeds. We found that trajectory error was significantly lower in the eight-bar mechanism (3.76 ± 0.18 cm, mean ± SD) than the four- (6.84 ± 0.37 cm) and six-bar (7.15 ± 0.48 cm) configurations (p < 0.0001). However, we did not observe differences in time-independent path error between configurations or in mechanical advantage (p > 0.05). We conclude that the optimized eight-bar configuration is best when the mechanism is tied to a constant velocity actuator, whereas the optimized four-bar mechanism performs as well if the actuator is well-controlled.

采用连杆机构的单自由度机械化步态训练器可以模仿人类使用者的步态轨迹，其成本可能低于当前的商业解决方案。我们的目标是从运动学的角度，即遵循自然步态模式的能力，评估最近推出的四杠，六杠和八杠设计。基于113位健康个体的步态数据库，我们合成了允许两次链接长度调整的机制，并比较了跨机制的踝关节位置的轨迹和路径准确性。我们使用交叉验证来获得与数据无关的准确性度量。我们使用了来自大型人类数据库的步态轨迹，并预测了其他轨迹来代表不同高度和速度的影响。p  <0.0001）。但是，我们没有观察到配置之间的时间无关路径误差或机械优势之间的差异（p  > 0.05）。我们得出的结论是，当该机构与恒速执行器相连时，最佳的八连杆机构配置是最佳的，而如果执行机构受到良好的控制，则最佳的四连杆机构也可以发挥最佳性能。