Floor Classification Algorithm Based on Stair Awareness Using MEMS IMU

The necessity of indoor navigation is increasing with the growing number of firemen who died in the line of duty at the scene of the rescue. The conventional radio navigation and localization using wireless signals would malfunction in disaster, so Pedestrian Dead Reckoning (PDR) is applied to the system. In real world situations, to determine current floor is significant for identifying the location of firefighters. In this paper, the floor classification algorithm is proposed using a MEMS IMU mounted on a shoe with stair awareness. On the basis of the IMU output, PDR system is used to decide current location with step detection, step length and azimuth estimation. As a pedestrian walks inside a building, stair motion is detected through Extended Kalman Filter (EKF) prior to floor division. The proposed algorithm is based on calculating heights through EKF using inertial sensor data and Zero velocity UPdaTe (ZUPT). Various types of stairs in buildings are largely categorized into several groups with the heading estimates from EKF. With the information of a user motion and the number of stairs, a current floor is determined based on the stair types. The experimental result of the proposed algorithm shows that floor level is classified regardless of stair types and the number of stair where a pedestrian moved.