Radar-based georeferencing system for estimation of changing river-shores

This paper proposes a radar-based sensing system to estimate the coordinates for river shores contributing to reduce the associated costs with the navigation-chart update process. The system comprises a commercially available radar, a GPS, and heading sensor that communicate to a central node that runs algorithms for shore estimation. In particular, to extract and deliver the radar information, a dedicated FPGA-based unit was implemented to interface with the radar internal signaling and to deliver the processed information to the main computer. In the central node, a novel mathematical model based on projection of parametric ellipsoids is programmed such that the shore estimation can be used without any geographical restriction. The estimation model uses a K-means algorithm to segment the sample data and applies a biased estimator to reduce the error of the output. The proposed system performance is validated using official geographical information showing that in fact a tradeoff between accuracy and cost is possible while maintaining acceptable shore estimation.