Development of a wireless displacement estimation system using IMU-based device

Abstract

Estimation of displacement is an information required for daily operation monitoring systems to monitor human health or to locate users in buildings, basements, tunnels and similar places which under the same conditions that the global positioning signal (GPS) level is from very weak to completely absent; and is the measurement technique by using multimetric data fusion. Most current displacement estimation methods require a lot of infrastructures and devices such as UWB, wifi access points, cameras. Hence, estimation methods that ultilize inertial measurement unit (IMU) and integrate acceleration to get diplacement are effective alternatives since the three-axis accelerometer embedded in IMU usually low cost, easy to adjust and low noise. The advantage of this approach is that the IMU-based device is compact, easy to install and put on user's body. However, these methods expose some weaknesses when used in large-scale indoor structures such as multi-storey buildings due to the need to compensate azimuth estimation which is drifted overtime and is employed for calculating displacement with refer to earth frame as a base station. This article proposes a low-cost wireless displacement estimation system developed with IMU. The system employs a Kalman-filter type in indirect form for orientation estimates and Median-filter algorithm for classification of motion modes. In order to verify the proposed system in terms of accuracy and feasibility, a device was designed in a wearable form and tetsted on a multi-storey building in university. The wearable device ultilized IMU model MPU9250 and results recored wirelessly via Xbee devices in order to test the system performance in such senarios as climbing/descending staircases only, climbing/descending staircases through one floor combined with walking. Experiments are repeated for root mean square error (RMSE) computation based on the ground-truth. The proposed system performance is evaluated accordingly to RMSE. The experimental results demonstrate RMSE of 3.56%, 1.43%, for climbing/descending staircases only, climbing/descending staircases one floor combined with walking, respectively.