F. Anaya Reyes, A. Narayan, D. Shuo, S. Han, S. Bhattacharya, H. Yu
National University of Singapore,
Singapore
Keywords: Stroke, Rehabilitation robotics, Gait recognition, Wearable sensors, Smart walker.
Summary:
Approximately two-thirds of stroke survivors have difficulty walking. Hemiplegia that often follows stroke is characterized by impaired balance, asymmetric voluntary movements, and severe muscle weakness or paralysis. Existing robotic platforms for ambulatory gait training typically provide balance and body weight support but lack mechanisms to improve gait speed. This is an important limitation because gait speed is the most common outcome measure for gait training after stroke. In addition, the ability of mobile gait platforms to provide stable body support trades off against mobility. Providing a stable support base for the patient's body usually requires a mechanical structure that is large and heavy. Consequently, the patient is subject to the inertial effects of the device. In this work, we present a novel mobile robotic platform with pelvic motion and trunk support for overground gait rehabilitation after stroke. The system consists of an omnidirectional platform with a body harness attached to it via a 6-axis force/torque sensor (FT). The platform has two rear actuated split-offset casters and two passive front wheels. The support frame for the FT sensor and harness is mounted on a linear drive to allow vertical movement of the trunk. We employ admittance control to compensate for the mass and dissipation effects of the moving robotic platform. In addition to its primary fall prevention function, our device combines two therapeutic interventions: forward propulsion of the trunk and partial body weight support (BWS). Forward trunk propulsion is performed with the help of a horizontal force that is modulated according to the patient's walking speed and turning rate to ensure easy adaptation. Robotic therapy is complemented by functional electrical stimulation (FES) to stimulate the ankle dorsiflexors during the swing phase. Lower limb kinematics and real-time gait event detection by inertial measurement units are performed to trigger the FES. In this way, the system takes into account natural variations in muscle response to stimulation. To verify the efficacy of the technology, we have conducted usability tests with patients who have suffered a stroke. We obtained evidence that the platform's assistance is effective in increasing self-selected gait cadence and speed, as well as improving temporal gait symmetries. The FES system, meanwhile, proved effective in compensating for joint motion deficits. At a minimum, the observed changes indicate that patients can adapt to movement in the walker without undue difficulty. These results suggest that the mobile platform may be an effective tool for gait training after stroke. However, the important research question is how much of the observed effects the patient will retain after completing a rehabilitation program with the walker. Future work includes a self-controlled study with multiple patients that should provide the answer to the retention question.