Location tracking using IMU and UWB

Abstract

This report includes research, development and testing of indoor positioning system using Inertial Measurement Unit (IMU) and Ultra-wideband (UWB). UWB can receive its location from fixed UWB anchors placed in the environment. However, it performs poorly in NLOS condition. On the other hand, IMU can measure its location without relying on external sensors. The drawback is that IMU sensors are prone to noises and biases, and they are not reliable for long-term use. Therefore, localization systems using IMU and UWB have become very popular. In this report, Mahony Filter is applied to IMU outputs to get the heading of the system. Calibrated accelerometer data from IMU is integrated twice to get displacement measurement, which is fused with position data from UWB using Extended Kalman Filter. Fused position data is then transferred to back-end computer via User Datagram Protocol (UDP) for data recording and visualization. The accuracy of positioning is tested by varying the update rate of UWB from 1 Hz to 10 Hz, and changing the way the device is carried during the tests. It was found that the system with UWB alone is accurate to minimum ±20cm when it is within the UWB anchor cluster. If it is away from the cluster more than 2m, the tag does not receive the position from anchors, and the accuracy drops drastically. However, positioning system using IMU and UWB can maintain the position error at ± 40-80cm at 10Hz. At 1Hz, the latency in getting estimate position increases due to UWB signal path obstruction, and accuracy drops to ± 1.5 m. More improvements are to be done to IMU to reduce the drift without affecting the double integral results when UWB cannot give its position.