Development and testing of a GPS-augmented multi-sensor vehicle navigationsystem

An integrated multi-sensor vehicle navigation system is presented that uses a low-cost rate gyro and differential odometry to supplement GPS under sig nal masking conditions such as tree foliage and urban canyons. Signal maski ng is often accompanied by extreme multi-path in urban centres with tall bu ildings, and is also found in wooded areas, enclosed car parks, tunnels, et c. The purpose of the system tested is to provide an accuracy of better tha n 20 metres almost 100% of the time throughout these interruptions, which a re assumed to last up to a few minutes. The equipment used is discussed in detail, as is the method used for filtering measurements. Results are prese nted from tests carried out in an urban core with relatively long periods o f signal loss - up to several minutes over a 6-km test circuit. Tests in ur ban canyons demonstrate that it is difficult to reach the above specificati ons with aiding from differential odometry alone due to the high precision of the wheel-scale factor required. However, with the use of a rate gyro an d odometry, RMS errors are below 20 metres while availability is nearly 100 %. Some of the large deviations could probably be better controlled if GPS multi-path errors were detected before they are allowed to corrupt the fil tered solution.