Real time, high accuracy, relative state estimation for multiple vehicle systems.

摘要

This dissertation presents the development, implementation, and test results from a new instrumentation package for relative navigation between moving vehicles. The instrumentation package on each vehicle is composed of a GPS (Global Positioning System) receiver, an IMU (Inertial Measurement Unit), a wireless communication system, and a modular computer system. The GPS places all vehicles into the same inertial reference frame and provides a common clock allowing synchronization among all instrument packages. The IMU tracks the high frequency motion of the vehicle alleviating the need for a fixed base station. The wireless communication system communicates GPS code and carrier phase measurements and computed state estimates from each vehicle at a rate fast enough to capture the dynamic changes in the vehicles. This data representing both GPS and IMU measurements from each vehicle is fused together on each vehicle to produce position, velocity and attitude estimates relative to the other vehicles. This capability to estimate relative motion without a base station appears unique. Furthermore, the application of fusion algorithms to address this new estimation problem is unique. The use of carrier phase provides very accurate relative measurements. In constructing carrier phase measurement, the integer number of wave lengths between vehicles must be resolved. Although there exist integer resolution schemes, these algorithms are ad hoe. The scheme presented here is based on generating the conditional probability of the hypothesis of each integer given the measurement sequence. This nonlinear filter is an elegant and novel contribution. The entire system is tested in real time in an experiment intended to validate the measurement accuracy. The system built using the algorithms designed in this dissertation is capable of estimating relative range to less than 5 cm. RMS, relative roll and pitch to less than 0.2 degrees RMS, and relative yaw to less than 0.7 degrees RMS. This system is designed and built for use in a formation flight drag reduction experiment. The system will be used on two F-18's at NASA Dryden to estimate the relative state between the vehicles and provide these estimates as an input to the formation flight control system.