Using Footsteps to Estimate Changes in the Desired Gait Speed of an Exoskeleton User
Using Footsteps to Estimate Changes in the Desired Gait Speed of an Exoskeleton User
复制标题
DOI：
10.1109/lra.2021.3096163
复制
发表时间：
2021-10-01
影响因子：
5.2
通讯作者：
Wensing, Patrick M.
中科院分区：
文献类型：
--
作者：
Karulkar, Roopak M.;Wensing, Patrick M.
这封信概述了一个估计框架，以检测步行过程中辅助外骨骼用户的预期步态速度的变化。提出了两次每步估计策略，称为受支持的卡尔曼过滤器，以利用步行位置的变化来推断所需速度的变化。估算器的第一阶段将贝叶斯更新在第二阶段传递给卡尔曼过滤器之前，将贝叶斯更新到大众状态中心。贝叶斯更新依赖于预测的步长之间的比较，该预测的步长计算为预期速度的函数和测量的步长长度，其差异提供了对用户意图的见解。该框架是通过在EKSO GT外骨骼中获得的传感器数据测试的，用于从中间到下脊柱的脊髓损伤（ISCIS）的用户（ISCIS）的用户。试验包括用户在指挥时改变步态速度。提出的框架能够在用户物理改变速度之前预测这些期望的更改。还发现，使用髋关节均方根（RMS）电流的测量值提高了估计量在预测ISCIS个体的意图变化方面的有效性。
This letter outlines an estimation framework to detect changes in the intended gait speed of an assistive exoskeleton user during walking. A twice-per-step estimation strategy, termed a Buttressed Kalman Filter, is presented to leverage changes in foot placement to infer changes in desired speed. The first stage of the estimator applies a Bayesian update to the center of mass state at midstance before it is passed to a Kalman filter in the second stage. The Bayesian update relies on the comparison between a predicted step length computed as a function of the intended velocity and the measured step length, the difference of which provides insight into user intent. This framework was tested with sensor data acquired from walking trials in an Ekso GT exoskeleton for users with and without incomplete Spinal Cord Injuries (iSCIs) from the middle to lower spine. The trials consisted of users changing their gait speed upon command. The presented framework was able to anticipate these desired changes before users physically changed their speed. It was also found that using measurements of the root mean square (RMS) current of the hip motors increased the effectiveness of the estimator in predicting intent changes for individuals with iSCIs.