Ultra-wideband (UWB) local positioning system (LPS) - An experimental study and analysis

An experimental study of recently developed UWB ranging radios is presented in this paper. The new radios measure ranges with a resolution of +/- 6 inches over a range of a kilometer, and can update at a rate of a thousand samples per second. However, measured ranges often have uncertain biases and occasional large sporadic large errors due to practical issues such as multipath and attenuation effect caused by radio penetrating through mediums. The paper will discuss experimental tests and results that characterize the ranging characteristics of the UWB radios in terms of statistics on accuracy and repeatability. Based on the observed characteristics, a fuzzy adaptive double Kalman filtering technique was developed to improve convergence to desired estimates. The double filtering technique consists of an overall filter that supervises a neighborhood filter via a fuzzy logic that provides initial states, minimizes large sporadic noise and recalibrates for biases. The paper will also present an ad hoc UWB local positioning system based on triangulation of the ranges. Statistical bounds for errors in the calculation will be analyzed and compared using techniques such as the Cramer-Rao bound, sensitivity analysis, and Monte Carlo simulation. Finally, sensor fusion the UWB location with vehicle kinematics/dynamics was implemented. The paper will present an experimental demonstration of autonomous mobile robot navigation using the UWB LPS.