Real evaluation for designing sensor fusion in UAV platforms

Abstract

Evaluation of UAV systems is mostly based on simulation tools that are manually configured to define the trajectory (ground truth trajectory) for comparing with the system output. In this work, the authors present an original method to evaluate the performance of UAV platform in real situations without considering simulations. The proposed evaluation methodology allows calculating the system accuracy and robustness with a considerable number of samples, accumulating the performance of different missions in the same conditions. The main innovation is an alternative evaluation for designing sensor fusion parameters using real performance indicators of navigation accuracy in UAVs based on a commercially available flight controller and peripherals. This methodology and selected performance indicators allow to select the best parameters for the fusion system of a determined configuration of sensors and a predefined real mission not requiring ground truth. The selected platform is described highlighting the available sensors and data processing software, and the experimental methodology is proposed to characterize the sensor data fusion output. The results show in detail the presented performance metrics for a set of trajectories in order to determine the best choice of parameters using quality measurements of navigation output.