Positioning Integrity, Availability and Precision for Journey Planning and Navigation Using GNSS Integrated with Low-Cost Sensors

Abstract

Intelligent transportation systems (ITS) and autonomous vehicles need accurate localization solutions for applications such as lane identification and collision avoidance. Accuracy at the sub-metre level with 100% availability and integrity above 99% are desired. At present, no single positioning sensor can meet these requirements. In this study, continuous positioning is obtained by using GNSS-RTK with MEMS IMU and vehicle odometer. For journey planning, a new approach is presented to choose the route with best positioning performance, where positioning integrity, availability and precision along different possible routes are predicted. Prediction of the satellites in view is performed using 3D city models. A method is presented for computation of the Protection Levels (PL), which bound the position errors along the direction of motion of the vehicle and for the cross-track direction. In the case of actual navigation, the Fault Detection and Exclusion (FDE) process in integrity monitoring (IM) includes applying the Chi-square test for detection of faults and identification is performed using the w-test, both are applied in the observation domain. The exclusion of observations is confirmed using the solution-separation method, performed in the position domain. A new test defined as ‘FIT” is presented, which checks the impact of the remaining measurement and system errors after the FDE on the final position. The proposed methods are demonstrated through a kinematic test run in an urban area, which represents a challenging environment for ITS applications. The predicted positioning metrics were compared with their values computed from the actual test data. Results show that the 3D map-aided method is able to reasonably predict satellite visibility, and accordingly aid prediction of the integrity, availability and precision. However, it was optimistic in some sections of the road. When using RTK, tight PLs were produced and an alert limit of 1.0 m can be implemented with IM availability of 99%. The use of IMU+vehicle odometer computes the time-change in positioning, hence, their biases accumulate with time and their use in bridging RTK positioning should be limited to short periods.