Ankle foot orthoses (AFOs) are widely used as assistive and/or rehabilitation devices to correct gait of people with lower leg neuromuscular dysfunction and muscle weakness. An AFO is an external device worn on the lower leg and foot that provides mechanical assistance at the ankle joint. Active AFOs are powered devices that provide assistive torque at the ankle joint. We have previously developed the Portable Powered Ankle-Foot Orthosis (PPAFO), which uses pneumatic power via compressed CO2 to provide untethered ankle torque assistance. My dissertation work focused on the development of control strategies for the PPAFO that are robust, applicable to different gait patterns, functional in different gait modes, and energy efficient. Three studies addressing these topics are presented in this dissertation:​ (1) estimation of the system state during the gait cycle for actuation control;​ (2) gait mode recognition and control (e.g., stair and ramp descent/ascent);​ and (3) system analysis and improvement of pneumatic energy efficiency.

Study 1 presents the work on estimating the gait state for powered AFO control. The proposed scheme is a state estimator that reliably detects gait events while using only a limited array of sensor data (ankle angle and contact forces at the toe and heel). Our approach uses cross-correlation between a window of past measurements and a learned model to estimate the configuration of the human walker, and detects gait events based on this estimate. The proposed state estimator was experimentally validated on five healthy subjects and with one subject that had neuromuscular impairment. The results highlight that this new approach reduced the root-mean-square error by up to 40% for the impaired subject and up to 49% for the healthy subjects compared to a simplistic direct event controller. Moreover, this approach was robust to perturbations due to changes in walking speed and control actuation.

Study 2 proposed a gait mode recognition and control solution to identify a change in walking environment such as stair and ramp ascent/descent. Since portability is a key to the success of the PPAFO as a gait assist device, it is critical to recognize and control for multiple gait modes (i.e., level walking, stair ascent/descent and ramp ascent/descent). While manual mode switching is implemented on most devices, we propose an automatic gait mode recognition scheme by tracking the 3D position of the PPAFO from an inertial measurement unit (IMU). Experimental results indicate that the controller was able to identify the position, orientation and gait mode in real time, and properly control the actuation. The overall recognition success rate was over 97%.

Study 3 addressed improving operational runtime by analyzing the system efficiency and proposing an energy harvesting and recycling scheme to save fuel. Through a systematic analysis, the overall system efficiency was determined by deriving both the system operational efficiency and the system component efficiency. An improved pneumatic operation utilized an accumulator to harvest and then recycle the exhaust energy from a previous actuation to power the subsequent actuation. The overall system efficiency was improved from 20.5% to 29.7%, a fuel savings of 31%. Work losses across pneumatic components and solutions to improve them were quantified and discussed.

Future work including reducing delay in recognition, exploring faulty recognition, additional options for harvesting human energy, and learning control were proposed.