Abstract:

The IMU (Inertial Measurement Unit) sensor is widely used to understand human motion. To better estimate the orientation of human segments, many filters have been developed using the concepts of Kalman filters and complementary filters. However, most of these filters are not designed for the movement of human body segments, which have rapidly changing dynamic characteristics and inherent impact potential. To address this weakness, an inertial orientation estimation filter that adaptively changes its parameters as it detects impacts is introduced. A real-time algorithm to identify heel strikes for use with this orientation estimation filter in walking is also proposed. The proposed filter was tested in constant and variable speed walking, which are common human motions in everyday life. It was also tested on gravel and sandy roads to see how it performs in outdoor environments. The proposed filter reduced the error more effectively than other filters in the literature. This filter, combined with other real-time impact detection algorithms, provides a framework that can be applied to a wide range of human motion.