An Integration Algorithm for Urban Kinematic Mapping: a Kalman Filter Solution Using a Modified Form of the GPS Triple Difference and Dead Reckoning Observables.

摘要

To improve kinematic mapping performance during periods of partial satellite visibility, this paper presents the algorithms developed to integrate GPS carrier phase, vehicle odometer and gyroscope rotation measurements to determine the position and orientation of a moving vehicle. Based around the standard Kalman filter algorithms, a measurement model has been derived which has a translational component and a rotational, or orientation component. The translational component continuously tracks the change in vehicle position over time using modified GPS triple difference carrier phase measurements and measurements of distance from the vehicle odometer to update predictions from a constant acceleration dynamic model. The rotational component tracks the orientation of the vehicle coordinate frame using measurements of orientation change provided by three mutually-orthogonal rate gyroscopes to update predictions from a constant angular acceleration dynamic model. Quaternions, also known as hypercomplex numbers are used as a novel means of tracking the orientation of the vehicle coordinate frame. This measurement model does not solve for any ambiguity terms and therefore does not attempt to provide absolute positioning accuracy, but relative positioning accuracy. It is therefore best suited to complement an existing kinematic GPS positioning system by bridging periods of inadequate satellite availability with a more accurate and robust solution than that provided by a stand alone, low-cost dead reckoning system. The algorithms developed, testing procedures adopted and results obtained from field tests conducted in Melbourne, Australia are fully described in this paper. Practical results and recommendations for future algorithm development will also be presented.