Adaptive trajectory segmentation method and its application in in-car navigation system.

摘要

The main purpose of a navigation system is to determine an appropriate position on a road segment represented in the database. This position is representative of the vehicle's position and is displayed as information to the driver. There are conflicting issues that a navigation system must resolve. A position has to be determined in real time using dead reckoning sensor measurements, GPS position, and the database with accuracy sufficient to guide the driver from the origin to the destination. But inexpensive sensors must be used to reduce the cost of a navigation system. Existing methods such as Kalman filtering and map matching are used to solve this problem but their implementations are not adequate enough. The purpose of this research is to investigate methods and algorithms that can take advantage of the properties of each kind of input data and meet the requirements.; A new method called adaptive trajectory segmentation (ATS) is implemented to continuously capture and derive vehicle trajectory into data sets of small lengths from dead reckoning measurements, GPS measurements, and the database. The derived data are conveniently used in more comprehensive methods and algorithms to determine a number of different dead reckoning sensor parameter. The same data is also used in simple and flexible methods of map matching for accurately determining segment position and correcting the dead reckoned position and heading. This process is done in such a way that the dead reckoned position and heading can be made to conform with the geometry of the segments. Methods are also used to first determine the quality of GPS data before using them to determine the dead reckoning sensor parameter and to correct the vehicle position when driven off-road. These methods and algorithms are tested using data collected from a number of test drives. The results show that the segment positions can be determined with an accuracy of 15 meters or better anywhere without continuous GPS availability.