A Continuous sEMG-Based Prosthetic Hand Control System Without Motion or Force Sensors

TL;DR

The paper addresses sensor burden in sEMG-based prosthetics by leveraging a near-linear mapping between $sEMG$ amplitudes and finger force $F$ to enable continuous control without motion or force sensors. It introduces the ResDD model with a linear backbone $G=WX$ and nonlinear correction $\sigma(G,X)=(WX)\circ X$, yielding $Y=G \circ X + G$, and uses two MVC-derived anchor points labeled +1 and -1 to define the mapping. Offline classification and interpolation analyses show ResDD captures the near-linear relationship and offers richer representations than a linear model while avoiding the instability of fully nonlinear networks; online sine-wave tracking and real-time control demonstrate practical usability and intuitive feedback. Together, these results suggest a sensor-free, accurate, and interpretable framework for sEMG-driven prosthetic hand control with potential for scalable clinical deployment.

Abstract

Regressively-based surface electromyography (sEMG) prosthetics are widely used for their ability to continuously convert muscle activity into finger force and motion. However, they typically require additional kinematic or dynamic sensors, which increases complexity and limits practical application. To address this, this paper proposes a method based on the simplified near-linear relationship between sEMG and finger force, using the near-linear model ResDD proposed in this work. By applying the principle that a line can be determined by two points, we eliminate the need for complex sensor calibration. Specifically, by recording the sEMG during maximum finger flexion and extension, and assigning corresponding forces of 1 and -1, the ResDD model can fit the simplified relationship between sEMG signals and force, enabling continuous prediction and control of finger force and gestures. Offline experiments were conducted to evaluate the model's classification accuracy and its ability to learn sufficient information. It uses interpolation analysis to open up the internal structure of the trained model and checks whether the fitted curve of the model conforms to the nearly linear relationship between sEMG and force. Finally, online control and sine wave tracking experiments were carried out to further verify the practicality of the proposed method. The results show that the method effectively extracts meaningful information from sEMG and accurately decodes them. The near-linear model sufficiently reflects the expected relationship between sEMG and finger force. Fitting this simplified near-linear relationship is adequate to achieve continuous and smooth control of finger force and gestures, confirming the feasibility and effectiveness of the proposed approach.

Figures (14)

• Figure 1: Idea and verification process
• Figure 2: Continuous prediction of finger force labels and virtual hands control system based on sEMG. Part A demonstrates the process of data acquisition, model training, and online application without the need for kinetic and kinematic sensors. Part B contrasts the differences in methods for gestures recognition and kinetic and kinematic information prediction between this study and previous research
• Figure 3: Relationship between force values of muscles and amplitude features values of sEMG
• Figure 4: List of 10 gestures in source gesture set and their force mode, participants hold each gesture for 30 seconds
• Figure 5: sEMG data were collected using a 12-channel sensor, covering as much of the forearm muscle groups as possible. The Unity 3D program was used to display guided gestures for participants in data collection