Dual-Purpose Co-Located Multimodal Sensing Gesture Recognition System: Intentional Resolution of Forearm and Wrist Muscle Activities

Accurate gesture recognition technology is of significance in human-machine interaction (HMI), such as prosthetic control. However, resolving intentional activities of distal upper limb musculature (forearm and wrist muscle groups) remains challenging due to limitations in biosignals. This study introduces a novel dual-purpose co-located multimodal sensing system integrating surface electromyography (sEMG), mechanomyography (MMG), and force myography (FMG), with a two-part architecture: a co-located multimodal sensing system for synchronized signal acquisition/transmission, and an intentional resolution gesture recognition system for gesture classification. Advancing conventional tri-modal approaches, it acquires complementary biosignals from identical anatomical sites and enables dual-position recognition (forearm and wrist) via optimized feature selection using analysis of variance (ANOVA)-based sequential forward selection (SFS) combined with classification models. The system was validated with 12 healthy participants performing 14 gestures (7 finger, 7 wrist) across both positions. Wilcoxon signed-rank tests (95% CIs) showed tri-modal fusion improved accuracy by 2.54–31.04% (vs. unimodal) and 1.11–12.11% (vs. bimodal), with all differences statistically significant (p ≤ 0.05). Average accuracy reached 97.1% (forearm) and 96.5% (wrist), demonstrating robust dual-position performance. These results substantiate the system’s effectiveness in enhancing intentional activity resolution and its feasibility for dual-purpose gesture recognition.