An optimal map-aided position estimator for tracking motor vehicles

Abstract

For motor vehicles, the road network represents a source of position information. The roads restrict the domain of the motor vehicle and therefore it follows that any measurement not falling within this domain is clearly affected by measurement noise. The authors have previously developed an estimator for translating position measurements onto a road thereby removing a component of the measurement noise. This work is now extended to cover the translation of velocity measurements and the joint estimation of position and velocity. Further to this the problem of applying the estimator to a complete road network has been analyzed. A Kalman filter, referred to as the spatially reduced Kalman filter (SRKF), has been developed to utilise the reduced dimensionality of the translated data. For each possible trajectory through the road network, an SRKF is initialised and updated. Target-tracking techniques have been adapted to enable the best trajectory at any given time to be selected and poor trajectories to be deleted. The result is an estimation process that results in better accuracy and effective road identification.