A Vehicle Map-matching Algorithm based on Measure Fuzzy Sorting

Abstract

The vehicle position obtained from GPS and dead reckoning is wildly applied to car navigation systems. However, the estimated position has an undesirable error due to the unknown GPS noise. To solve this problem, previous papers presented a method called map-matching to correct the position error. In this paper, we proposes a fuzzy ranking map matching algorithm based on measure factor. Comparing with other four algorithms, our algorithm improves in strategies of the error region determination, the road grid index and auto-adapted fuzzy sorting. To be specific, the error rectangle is firstly replaced by the error ellipse to reduce geometrical operation. Secondly, the grid index is adopted to accelerate the speed of filtering candidate road. At last, the relativity function and fuzzy sorting method help to sort the membership degree and to decide the matching road section. For the experiments, we implement a vehicle navigation system of five kinds of vehicle running status to testify the robustness and efficiency of this algorithm. The result shows that 96.7% of the GPS points are matched. In comparison with other algorithms, this algorithm had highest accuracy, which is of importance for vehicle navigation.