Walk distance estimation using an ankle-mounted inertial measurement unit

Abstract

This work proposes walk distance estimation methods suitable for low power, low computational capability devices, using an ankle-mounted inertial measurement unit. A stride detection method using gyroscope data was implemented, and two stride length estimation methods were developed using the stride cycle information: a simple method, which estimates the leg angle during the forward swing of the leg; and an improved method, which uses the inverted pendulum model to provide the initial conditions for integration of the gyroscope and accelerometer signals in the three-dimensional space. The proposed methods were compared with a two-dimensional stride length estimation method, highlighting the importance of misalignments during sensor placement. Compared to the two-dimensional method, the simple method proposed in this paper achieved approximately the same level of performance with lower computational costs, whereas the three-dimensional method achieved 67% to 78% improvement in performance.