Wrist torque estimation during simultaneous and continuously changing movements: surface vs. untargeted intramuscular EMG

This data was recorded for emg based force/Torque estimation. EMG and torque signals were collected during simultaneous, isometric, but continuously varying contractions, corresponding to two wrist DoF. The experiment was carried out in two trials with a 5-min rest in between. Each trial included six combinations of tasks, separated by 2 min of rest to minimize the effect of fatigue. The performed tasks were categorized into individual and combined (simultaneous) DoF to test the ability to estimate isolated torque and torque in two simultaneous DoF. The selected DoF were wrist flexion/extension (DoF1) and wrist supination/pronation (DoF2). The following six tasks were chosen: wrist flexion/extension, supination/pronation, simultaneous flexion + pronation/extension + supination. We chose to use only the two DoF, because no commercially available prosthetic device is currently capable of ulnar and radial deviation. All tasks were dynamically varying, following a sinusoidal profile lasting 30 s at maximum amplitudes of 3 Nm and 2 Nm for men and women, respectively. Torque level was limited to 2 Nm for female subjects, because concerns were raised during pilot testing about the combined movements being uncomfortable for higher torque levels. On the contrary, 2 Nm was too easy for most male subjects, thus 3 Nm was adopted to allow similar EMG activation for female subjects. The first and last 3 s of each profile consisted of a rest period. Sinusoidal profiles were used to elicit dynamic contractions, requiring constant changes in the torque (and EMG intensity) level of the performed task. The subjects were seated in a chair with their right arm placed in an arm rest, while the left arm was left relaxed on the table. Subjects were asked to track one profile or two simultaneous profiles, depending on whether it was a single DoF or two simultaneous DoF,The order of the tasks was randomized, and subjects received visual feedback as to how closely they tracked the profiles. The subjects were given sufficient time for training and to become familiar with the profiles. The frequency of the profiles was 0.5 Hz.

本数据集用于基于肌电图（Electromyography, EMG）的力/扭矩估计研究。实验在同步等长但持续变化的肌肉收缩过程中，采集了肌电与扭矩信号，对应腕部两个自由度（Degree of Freedom, DoF）运动。实验分两轮开展，两轮间隔5分钟休息时间。每轮测试包含6组任务，每组任务间以2分钟休息间隔，以尽可能降低疲劳效应的影响。所设置的任务分为单自由度与组合（同步）自由度两类，以评估模型对单自由度孤立扭矩以及双同步自由度联合扭矩的估计性能。本次实验选取的两个腕部自由度分别为：腕屈伸（自由度1，DoF1）与腕旋前/旋后（自由度2，DoF2）。本次选定的6组任务如下：腕部屈伸、腕部旋前/旋后、同步屈曲+旋前、同步屈曲+旋后、同步伸展+旋前以及同步伸展+旋后。本次实验仅选取这两个腕部自由度，原因在于当前尚无商业化假肢设备可实现腕部尺偏与桡偏运动。所有任务均为动态变化模式，遵循时长30秒的正弦轨迹，男性受试者的最大扭矩幅值为3牛米（Newton meter, Nm），女性为2牛米。女性受试者的扭矩上限被设定为2牛米，原因是预实验阶段发现，较高扭矩下的联合运动可能引发受试者不适。与之相反，2牛米的扭矩对于多数男性受试者而言难度偏低，因此将男性扭矩上限设为3牛米，以保证男女受试者的肌电激活水平相近。每条正弦轨迹的首尾3秒均设置为休息时段。采用正弦轨迹的目的是诱发动态肌肉收缩，使得任务过程中的扭矩（及肌电信号强度）持续动态变化。受试者端坐于实验座椅上，右臂置于臂托中，左臂自然放松放置于桌面。受试者需根据任务类型，追踪单条或两条同步的正弦轨迹：单自由度任务追踪单条轨迹，双同步自由度任务则追踪两条同步轨迹。任务执行顺序采用随机化设计，受试者可通过视觉反馈了解自身对轨迹的追踪精度。实验为受试者提供了充足的训练时间，使其熟悉各轨迹模式。轨迹的频率为0.5赫兹（Hertz, Hz）。