An ANN Based Human Walking Distance Estimation with an Inertial Measurement Unit

Estimation of the walking distance including stride length of a pedestrian is an essential part of a motion tracking and monitoring devices used in applications such as health care, search and rescue and games. One popular approach is to derive it through processing the acceleration signals output from the inertia measurement unit (IMU) attached to a foot or another part of the body of the pedestrian. This approach is well known for its large accumulative errors in integration of the signals with noises of a sensor and the difficult to accurately determine the time duration of a stride. This paper presents a novel approach to estimate walking distance from IMU sensor readings based on an artificial neural network (ANN). Experiments show that the proposed approach is more accurate and effective than traditional methods relying on double integral method.