Shank-RIO: A Shank-Mounted Ranging-Inertial Odometry for Gait Analysis and Positioning in Complex Environment

Abstract

Human gait parameter estimation and localization are very important for lower limb exoskeleton control, biped robots, and biomechanics research. Gait analysis can be performed by wearable sensors worn on the lower limb joints. However, these sensors are incapable of simultaneous gait event detection and positioning and have problems with heading drift and low sampling rates. In order to address these limitations and fulfill the requirements of gait analysis, we propose a multisensor fusion technique called shank-mounted ranging-inertial odometry (Shank-RIO). Shank-RIO utilizes the fusion of a LiDAR ranging sensor and an inertial sensor placed on the shank to provide trajectory estimations for the feet and the body. A modified linear inverted pendulum model (LIPM) is employed to estimate gait parameters. Moreover, the heading rotation of the body is determined by analyzing the curvature of the body trajectory using step length and width. Based on these estimations, gait events, gait parameters, and body position can be estimated. We conduct experiments in various complex indoor and outdoor scenarios to evaluate the performance of Shank-RIO. The RMSE of the step length of the Shank-RIO was 2.26 cm. The positioning accuracy and positioning error within 1000 m are 0.799% and 0.395%, respectively. The results demonstrate that Shank-RIO serves as an effective tool for gait analysis research.