Range-Only UWB Localization in Measurement of Cadastral Boundary Points

Cadastral boundary survey is an important section of immovable property registration which is heavily relied on manual operation and time-consuming. To reduce the work load of land area survey, a UWB-based high accuracy localization system is proposed and a neural network estimation model is applied to the system in this paper. The time-of-flight (TOF) method is used to obtain distances between four anchors and one moving tag. Three different combinations of activation function in ANN model are trained via keras to search the best result. With four anchors, the best neural network architecture reduces the error of x-axis by 79.00% and the error of y-axis by 37.25%. The proposed method can be used in accurate land area measurement with range of 10m.