Step Length Estimation Using Inertial Measurements Units

This paper focuses on step length estimation using either one inertial measurement unit (IMU) on each shank or one IMU each on the shank and thigh of a human subject. An observer design problem that requires estimating shank angle, thigh angle and bias parameters of the initial sensors in order to estimate step lengths is formulated. A nonlinear observer is designed using Lyapunov analysis to solve the formulated problem and utilizes an LMI to find a stabilizing observer gain. Experimental results are presented on the performance of the nonlinear observer and compared with reference measurements from an infrared camera capture system. The estimation system that utilizes only one inertial sensor on each shank is seen to provide an average step length accuracy of 91.21% while the estimation system that utilizes two sensors on each leg provides an average step length accuracy of 95.37%.