Accurate angle estimation in smart knee prostheses via magnetic implantable and skin-mounted sensors

In this work we investigated how to measure concurrently flexion-extension and internal-external rotations in a smart knee prosthesis. A configuration of magnetic sensors and magnets were designed and embedded in knee prostheses in which each sensor measures a mixture of information related to both rotations. Using correlation analyses, angle estimators were designed to separate the flexion-extension and internal-external rotations information. The estimators were validated in a mechanical knee simulator towards a reference system. The effect of imposed abduction-adduction was also analyzed on the estimations performances. To reduce the power consumption of the internal system, we reduced the sampling rate and duty cycled the sensors and compensated the lack of information with skin-mounted sensors on four subjects. The fusion between implantable and skin-mounted sensors drastically improved the flexion-extension angle estimation, but not the internal-external estimation.