Phase-Based Grasp Classification for Prosthetic Hand Control Using sEMG

Abstract

Pattern recognition using surface Electromyography (sEMG) applied on prosthesis control has attracted much attention in these years. In most of the existing methods, the sEMG signal during the firmly grasped period is used for grasp classification because good performance can be achieved due to its relatively stable signal. However, using the only the firmly grasped period may cause a delay to control the prosthetic hand gestures. Regarding this issue, we explored how grasp classification accuracy changes during the reaching and grasping process, and identified the period that can leverage the grasp classification accuracy and the earlier grasp detection. We found that the grasp classification accuracy increased along the hand gradually grasping the object till firmly grasped, and there is a sweet period before firmly grasped period, which could be suitable for early grasp classification with reduced delay. On top of this, we also explored corresponding training strategies for better grasp classification in real-time applications.

Integrating computer vision to prosthetic hand control with sEMG: Preliminary results in grasp classification

“…In our previous research (Wang et al, 2022), we achieved the best grasp classification outcome using the sEMG sweet period between 1,100 ms and 1,400 ms in the early grasping phase (pink zone in Figure 3). The sEMG sweet period was 800 ms behind the vision sweet period.…”

“…According to the research of Mason et al (2002) and Supuk et al (2011), the reaching phase starts with hands lifting and ends at touching the object, the early grasping phase starts with the object touching and ends by firmly holding the object, in the firm grasping phase, the force from hand remains steady to the end of grasp. The details of these three phases were discussed in our previous research (Wang et al, 2022). According to the segmentation for this dataset in our previous research, the reaching phase was from 0 to 1,020 ms, the early grasping phase was from 1,020 ms to 1,604 ms, and the firm grasping phase was from 1,604 ms to 4,480 ms. A phase example can be found in Figure 2.…”

“…Our last comparison of grasp classification over three methods focused on the sweet periods. Specifically, we calculated the mean accuracy of object and gesture recognition using data collected in their sweet periods and compared it to classification by sEMG from the best strategy we have achieved in the previous report (Wang et al, 2022). The results are shown in Figure 6, in which the object and gesture classification accuracy was calculated from valid frames in the sweet period.…”

“…Once the sweet period is identified, grasp classification by the camera can be automatically conducted during this interval without purposed confirmation. In our recent publication (Wang et al, 2022), a similar sweet period (for sEMG) right before the hand grasps the target object was identified for hand grasp classification by sEMG. It will be interesting to explore the vision sweet period again during the reach-and-grasp process and utilize both of them for better prosthetic hand control.…”

Intelligent Biosignal Processing in Wearable and Implantable Sensors

“…Jiang et al [ 12 ] have analyzed surface Electromyography (sEMG) and used it for prosthesis control. They explore how the grasp classification accuracy changes during reaching and grasping, and they identify the period during which the grasp classification accuracy and detection are high.…”