Design and Development of a Wearable Ankle Exoskeleton to Avert Freezing of Gait in Persons with Parkinson's Disease

Abstract

Freezing of gait (FoG) is a debilitating motor symptom of Parkinson’s disease (PD) characterized by sudden interruptions that increase the risk of falling. While exoskeletons have improved gait in other neurological populations, their application to PD remains limited. To date, there exists only one hip exoskeleton study to avert FoG. However, the ankle plays a central role in gait propulsion, making it a plausible and unexplored contributor to FoG. This study explores the gap by proposing a self-contained ankle exoskeleton to mitigate FoG, driven by an ankle-coaxial, quasi-direct drive motor and controlled by a temporal convolutional network (TCN) informed by inertial measurement units (IMUs) on the shank and foot. The TCN was pre-trained using IMU data from two publicly available datasets and fine-tuned using self-collected data from one able-bodied participant across ten level-ground walking trials of ~150 seconds each. The participant walked up and down a hall at a cadence of 90 steps per minute, pausing to stand at the startpoint, midpoint, and endpoint. Mean peak torque was 20.06 Nm at 55% of gait phase. The controller was more consistent at 110 steps/min (SD 0.35 Nm vs. 0.62 Nm at 100) but more accurate at 100 steps/min (5% timing lag vs. ~10% at 110), consistent with the TCN being fine-tuned at 90 steps/min. Initial development and validation of the design was limited to able-bodied walking trials, and future studies will use a bilateral system to determine its efficacy in averting FoG in PD patients.