Design of an infrastructureless in-door localization device using an IMU sensor

There has been an increasing demand for localization for personnel like firemen, and soldiers for various reasons ranging from safety to strategy planning and coordination. There is also a need for the localization system to be free from infrastructure. For example, it is not practical to place various transmitters in a building before the users enter the building. Many of the current methods involving inertial measurement unit (IMU) utilize step detection and step counting to estimate the displacement. This does not account for the various legs length and step sizes though. Some groups have proposed algorithm that involves placing the IMU on the foot to estimate the step size. However, users have commented that it affects their walking. Hence, this paper presents a new method to estimate both the forward displacement and orientation. In this paper, the sensor unit is placed at the pedestrians ankles for greater ease of usage. The 2D displacement is then computed based on the estimations of pitch angle, yaw angle and pedestrians leg length. The advantage of this method is that the pedestrians leg length is automatically estimated during walking by exploiting the motion equation of a simple pendulum model and hence, no prior measurement or training is required. The proposed method also employs the quaternion-based indirect Kalman filter to estimate the Euler angles containing the yaw angle (heading), the pitch angle and the roll angle. The heading (yaw angle) is corrected by updating the reading data of magnetometer an estimated magnetometer bias. The real-time localization system has been implemented and experiments involving various subjects are conducted. The experimental results demonstrates the accuracy with the averaged displacement error less than 3%.