Precise Probabilistic Grid Positioning based on GNSS Pseudoranges and Digital Elevation Models

Future applications for connected and automated driving depend on high-precision, lane selective positioning especially in dense urban environments. Estimating global user position is oftentimes based on Global Navigation Satellite Systems (GNSS), but stand-alone GNSS positioning methods do not meet the necessary requirements in terms of accuracy and robustness. To achieve higher accuracies, additional sensor information, e.g. digital maps (DM), is usually incorporated. Baseline state of the art GNSS positioning is based on deterministic and probabilistic estimation methods which do not integrate a-priori map data in a tightly coupled manner but rather perform a map matching after the GNSS position was estimated. The work presented in this paper provides a proof of concept of a novel likelihood based snapshot Probability Grid Positioning (PGP) approach for low-cost GNSS receivers using a digital elevation model (DEM). The proposed method is described in detail and validated in a real word, dynamic measurement scenario. The presented approach is compared with a conventional positioning method.