Abdominal massage for constipation has problems such as high labor costs, insufficient standardization, and large differences in efficacy. This study aims to develop a new type of abdominal massage robot to break through the bottlenecks of manual abdominal massage. This paper constructs a new six-degree-of-freedom multi-functional abdominal massage robot system, improves and designs a new series-parallel hybrid mechanical structure, and based on this, establishes an abdominal interaction force control strategy that uses a variable-stiffness physical model to represent the biomechanical properties of the abdomen. Verified by simulation and human experiments, this robot can accurately reproduce a variety of massage techniques, and demonstrate stable force control performance in prosthesis experiments and human abdominal tests, with the interaction force error controlled within 0.8 N, providing an innovative path for the intelligentization and standardization of clinical abdominal massage.

This research focuses on the crucial issue of ascertaining safe and suitable massage forces for abdominal massage robots. It does so by constructing variable stiffness soft tissue physical models and devising a machine learning-based prediction algorithm. Twelve healthy volunteers with diverse body types (underweight, standard, overweight) were enlisted. A self-developed experimental platform was utilized to gather high-precision mechanical data. Two physical models were formulated to depict the nonlinear force-displacement relationship of abdominal soft tissues. Model 1 was based on an exponential function, and Model 2 was based on a power function. Model 1 showed greater accuracy for standard body types (R² > 0.95 at 72.22% of data points) and underweight body types (R² > 0.95 at 100% of data points). In contrast, Model 2 was a better fit for overweight individuals (R² > 0.95 at 47.06% of data points). Furthermore, a transformer-based machine learning algorithm was developed. This algorithm predicts interaction forces using anthropometric and physiological characteristics. It achieved an accuracy of 92.60% within a pressing depth of 0–15 mm. In most situations, especially in full-range predictions (with an accuracy of 58.60%), the algorithm outperformed both physical models. This study presents a dual framework that combines biomechanical modeling and data-driven algorithms. It lays the theoretical and practical groundwork for safe and adaptable force control in abdominal massage robotics.

Surface electromyography (sEMG) provides a non-invasive measure of the neural drive transmitted from the central nervous system to muscles by capturing the spatiotemporal summation of motor unit action potentials at the skin surface, and is therefore widely used to study neuromuscular coordination during motor tasks. By reflecting neural drive transmitted from the central nervous system to peripheral muscles, sEMG provides valuable insights for investigating neuromuscular coordination during upper-limb motor tasks. Within the framework of modular motor control, muscle synergy analysis has been increasingly applied to characterize coordinated muscle activation patterns extracted from multi-channel sEMG recordings. In this study, sEMG signals were collected from twelve stroke patients and nine healthy subjects during robot-assisted upper-limb training, involving two movement trajectories (straight and rectangular) and multiple robot-assisted levels. Muscle synergies were extracted using non-negative matrix factorization (NMF). A synergy merging–splitting model, combined with a Functional Driving Ratio (FDR), was employed to characterize both the muscle synergy reorganization and the relative activation contributions of driving versus stabilizing muscle components in terms of motor control strategy. The results showed that healthy subjects maintained consistent muscle coordination patterns across different assistive levels, while making task-dependent adjustments to muscle activation to adapt to variations in movement trajectories. For stroke patients, higher functional status was correlated with more differentiated coordination patterns and relatively higher FDR values, suggesting greater reliance on task-relevant agonist muscles during movement execution. In contrast, lower-function patients exhibited less differentiated coordination patterns accompanied by reduced FDR values, indicating the increased involvement of stabilizing or antagonist muscles. This shift may reflect compensatory control strategies and the reduced efficiency of neuromuscular coordination during assisted upper-limb movements. These findings suggest that sEMG-based muscle synergy features and the FDR may provide quantitative, sensor-derived support for characterizing neuromuscular coordination during robot-assisted rehabilitation.