Velocity estimation for UAVs using ultra wide-band system

Abstract

In this paper, a 3D velocity estimation solution based on ultra wide-band (UWB) signals is developed for navigation of unmanned aerial vehicles (UAVs) in indoor spaces. The solution incorporates two consecutive gradient descent optimizations for position and velocity estimation, as well as two linear Kalman filters for removing noise on the range measurements and the vertical position estimate. The solution is verified by three experimental scenarios with manual and autonomous flights of a quadrotor. In our implementation, 8 anchors are employed to construct the UWB system, while the Vicon MoCap system is utilized as the ground-truth for assessing the performance of the solution. All proposed algorithms are implemented on a computer on-board the quadrotor, with the low-level control for the quadrotor provided by the PX4 platform. Compared to a motion capture system, a relatively cheap solution for both absolute position and velocity estimation is achieved using the UWB system, with sufficient accuracy for autonomous flight of UAVs in indoor areas.