Virtual Constraint Control of Powered Prosthetic Legs: Unifying the Gait Cycle

The lower limb amputee population is gradually increasing, primarily due to complications from vascular diseases. The vast majority of lower limb amputees use mechanically passive prosthetic legs, which are unable to provide energy input at the joints and can only dissipate energy during locomotion. To improve amputee gait, powered prosthetic legs are in development. Several control methods have been proposed for these devices, but almost all of them divide the gait cycle into multiple, sequential periods with different controllers. This results in many patient-specific control parameters and switching rules that must be tuned for a specific ambulation mode, such as a desired walking speed or slope. The different periods of gait could potentially be unified over the entire gait cycle by virtual kinematic constraints that are enforced using a torque control scheme. The prosthetic control method proposed as part of this dissertation work unifies the different periods of gait through virtual constraints that are driven by a human-inspired phase variable. A phase variable is a kinematic quantity corresponding to an unactuated degree of freedom that evolves monotonically during steady walking, thus representing the progression through the gait cycle. The unified controller was designed systematically by method of virtual constraints, which was implemented within an amputee biped walker model for different walking speeds. To validate this control strategy even further, a powered knee-ankle prosthesis was designed and built during the course of this dissertation work for experimental validation. The mechanical design and real-time control of the powered prosthesis is presented. Experiments were conducted with multiple above-knee amputee subjects walking across various speeds and inclines, while no control parameters were tuned. This verified that our unified control scheme can work seamlessly and efficiently for multiple amputee users, and also, for different ambulation modes without retuning the controller. Furthermore, this work has taken a step forward to providing a solution of the technical challenges for powered knee-ankle prostheses to be used in a clinical setting. An intuitive clinical user interface was developed for clinicians to change the prosthesis control based on their clinical insight and expertise. We performed a case study with a clinician adjusting the virtual constraint design on the prosthesis, which resulted in improvement of the amputee’s gait symmetry using our control strategy.