Map Matching in Road Crossings of Urban Canyons Based on Road Traverses and Linear Heading-Change Model

摘要

The objective of this paper is to find a robust and sustainable positioning approach in Hong Kong, where the Global Positioning System (GPS) signal is frequently blocked by high buildings, traffic tunnels, etc. After evaluating several kinds of complementary positioning means, the GPS/dead-reckoning (DR)/map-matching (MM) integrated positioning technology is selected. During navigation, if the GPS signal is good enough, the GPS/DR/MM mode will be used; otherwise, the DR/MM mode will be used, and the MM algorithm will correct the DR system in a timely fashion. Generally, if the vehicle is running in a straight road, heading-related parameters in the DR system will be corrected, and if the vehicle makes a turn, both distance- and heading-related parameters will be corrected. Unfortunately, existing MM algorithms show poor performance near a road crossing or in a dense road network. To resolve this problem, based on road traverses and a linear heading-change model, an MM algorithm that considers historical information is proposed. First, a unified point-based MM calculating framework is presented, and the disadvantages of several kinds of existing MM algorithms are also given. After that, a single-line–single-direction road network model is presented, and the MM algorithm is always executed on road traverses instead of a single road so that historical information can be easily kept. By adopting the linear heading-change model, the problem of no analytical solution in the MM that considers heading information is obviated, and the MM ambiguities in road crossings can be greatly removed. Finally, an onsite MM test in Hong Kong is presented, and the test proves that the proposed MM algorithm provides long-time robust positioning performance.