Comparison of two sensor data fusion methods in a tightly coupled UWB/IMU 3-D localization system

There currently exists a high demand for vehicles with ever increasing levels of autonomy. While some of these vehicles may depend on GPS for localization, other vehicles will require a more precise localization solution in order to perform their tasks. In addition, some vehicles may need to work in GPS denied environments. These issues could be addressed with the use of an Ultra-Wide Band (UWB) ranging sensor fused with an Inertial-Measurement-Unit (IMU) using an Extended Kalman Filter (EKF). The main goal of this work is to investigate and compare two different sensor data fusion techniques to incorporate a 3-axis 9-DOF IMU in a tightly coupled fashion to a 3-D positioning solution that is derived from UWB signals. The techniques differ from each other in that the first fusion of the IMU data occurs in the prediction step and the second fusion occurs in the update step. Experimental results obtained with a quadcopter show that the data fusion performed in the update step outperforms the fusion performed in the prediction step. The results also show that when using very accurate UWB ranging sensors, the use of IMU data does not improve significantly the accuracy of the position. However, the integration of IMU data in the update step increases the robustness of the EKF against erroneous modelling of the process noise.