Investigation and Modeling for Mechanism of Human Slippage Sensation

在本研究中，我们尝试将人类灵巧的抓取应用到机器人中。因此，我们将重点放在与触觉感知密切相关的人类滑动受体上。首先，为了测量人体对滑移的感知能力，我们研制了一种滑移力表示机，能够在手指上显示各种滑移力。同时，我们采用PEST(参数估计顺序测试)方法作为心理物理实验，因为PEST方法能够有效地获得阈值和主观等价性，并基于PEST方法开发了计算机程序来控制现有的机器。在一系列实验中，我们采用滑移力2、2.75、3.5、4.25和5N作为PEST方法的标准刺激。对于光滑的亚克力板材表面，较小的滑移力会产生较大的韦伯分数(阈值与标准刺激之比)0.14，且韦伯分数随标准刺激的增加而增大。这意味着滑移力的检测精度随着滑移力的减小而降低。与视觉和听觉等其他感觉相比，这一结果是相当正常的现象。接下来，我们考察了表面条件对滑移力阈值的影响。砂纸表面的弯曲门槛比亚克力表面要小。这是因为，由于需要更大的抓地力来防止受试者的手指在表面上滑动，因此它为人类滑动感受器提供了噪音刺激。最后对抓取力与肌电输出之间的关系进行了检验，相关系数均在0.9以上。说明肌电可以作为抓持力传感器。

In the present study, we attempted to apply the human dextrous grasping to robotics. Due to this, we are focusing on human slippage receptor relating closely to haptic sensing. First of all, in order to measure human capability of slippage sensing, we developed a slippage force presentation machine capable of presenting various slippage force on human fingers. Also, we adopted PEST (Parameter Estimation Sequential Testing) method as psychophysical experiment because it was able to obtain efficiently threshold and subjective equivalence and developed computer programs to control the present machine based on PEST method. In a series of experiments, we adopted slippage forces 2, 2.75, 3.5, 4.25 and 5 N as standard stimuli for PEST method. For smooth surface of acrylic board, rather small slippage force caused slight large Weber fraction (ratio of threshold and standard stimulus) 0.14 and Weber fraction was increasing with increasing standard stimulus. It means that detecting precision for slippage force is decreasing with decreasing slippage force. This result is quite normal phenomenon comparing with other sensations such as vision and hearing. Next, we examined influence of surface condition on threshold of slippage force. Abrasive paper surface coursed threshold to be smaller comparing with acrylic surface. This is because, since higher grasping force is needed to prevent subject fingers from slipping on the surface in the case of acrylic surface, it provides noise stimulus for human slippage receptors. Finally, we examined relationship between grasping force and output of EMG (electromyrogram) to obtain high correlation coefficient exceeding 0.9. It indicates that we can use EMG as a grasping force sensor.