How does the ankle complex respond to cable-driven assistance spanning the talocrural and subtalar joints during walking?

Cable-driven ankle exoskeletons are primarily designed to assist plantarflexion, but their actuation cables also span the subtalar joint, potentially producing unintended inversion-eversion torques. These unintended torques can affect frontal-plane kinematics, joint coordination, gait stability, and assistance efficiency. This study investigated how the ankle complex responds to multi-dimensional assistance torques during walking. To this end, we established an assistive torque model based on anatomical joint axes and developed a dual-cable-driven ankle exoskeleton for evaluation. Four assistance modes were examined: three single-cable modes applying force from medial (Med), central (Mid), or lateral (Lat) heel positions, and a dual-cable biomimetic (Bionic) mode replicating physiological torque distribution. Six healthy participants performed treadmill walking trials in each mode, with multiple biomechanical and physiological variables recorded. Simulation and experimental results showed that the Lat mode generated eversion moments, increased eversion, shifted the center of pressure (CoP) medially, and reduced mediolateral center-of-mass sway by 9% compared with unassisted walking, thereby improving stability. The Med and Mid modes induced inversion moments, increased inversion, and shifted the CoP laterally, with Med producing the strongest effect. Compared with the other modes, the Bionic mode achieved the greatest reduction in muscle activation relative to unassisted walking, lowering soleus EMG by 10% and that of the ankle inversion muscle group and peroneus brevis by 18-22%, while best preserving ankle coordination patterns. These findings highlight the importance of frontal-plane dynamics in ankle exoskeleton control and the benefits of biomimetic torque distribution in preserving coordination and reducing locomotor effort in assisted walking.