Continuous Environment Mapping for Enhanced Low-cost Urban Navigation

Urban environments introduce a challenge for unaided GNSS to provide accurate coordinates at all times, which makes this approach insufficient for many applications. Recent developments in intelligent transportation systems lead to an evolution of such techniques as map matching, 3D map aiding, shadow matching, and visual-aided navigation. These methods along with inertial-sensor integration proved to improve the capability of low-cost-equipment-based precise positioning applications. The plethora of available sensors allows devices to collect rich contextual data and permits the user to adapt the positioning engine parameters to different environments. The limitation of past research using a single positioning platform is that context detection is solved independently, based on self-monitoring. Individual device sensor readings are prone to being affected by random errors and to increase the reliability of context detection, multiple readings should be made, which is not feasible in many real-time applications. In an urban setting, it is reasonable to assume that many identical platforms will collect similar sensor readings when operating in a given area. The research described in this paper aims to explore the use of collectively recorded context information with further processing, integrating the data with a map and feeding the summarized results to a user. A two-week piloting dataset in a dense urban environment was collected with two low-cost positioning boards with the same hardware and firmware configuration. It was shown in previous research that the signal-to-noise ratio measurement is a sufficient metric to distinguish not only between indoor and outdoor environments, but to detect signal jamming events and to characterise the quality of the GNSS signal reception in the area. The uniformity of the configuration of the devices allowed for reliable spatial mapping of the signal-to-noise-ratio measurements. The novel contributions of this paper are as follows. First, the implemented architecture capable of continuous real-time context mapping is described. Second, a context map-aided measurements noise adjustment algorithm is designed. Third, the benefits of the context-database-aiding are demonstrated in the case of pure GNSS coordinate determination.