Vehicles driving in urban canyons are always confronted with a degraded Global Navigation Satellite System (GNSS) signal environment. The surrounding obstacles may cause signal reflections or blockages, which lead to large multipath noises and intermittent GNSS reception. Under these circumstances, it is not feasible to use conventional real-time kinematic (RTK) algorithms to maintain high-precision performance for positioning. In order to meet the special requirements of safety-critical applications under non-ideal observation conditions, a novel tightly coupled RTK/Inertial Navigation System (INS) algorithm is proposed in this paper, which can provide accurate and reliable positioning results continuously. Our integrated RTK/INS algorithm has three features. Firstly, INS measurements are used to help search for integer ambiguities in the position domain. INS solutions can provide a more accurate initial location and a more efficient search region. Secondly, the criterion for determining whether a candidate position is the correct solution is only related to the fractional value of the carrier-phase measurement. Thus, the new algorithm is immune to cycle slips as well as large pseudorange noises. Thirdly, our algorithm can provide more accurate ranging information than the pseudorange, even though it may not necessarily be fixed successfully, because it selects the weighted ambiguity solution as the result rather than the candidate point with maximum probability. The proposed algorithm is demonstrated on both simulated and real datasets. Compared with single epoch RTK and conventional tightly coupled RTK/INS integrations that search integer ambiguities in the ambiguity domain, our method attains better accuracy and stability in a simulated environment. Moreover, the real experimental results are presented to validate the performance of the new integrated navigation algorithm.
Keywords: INS; RTK; high-precision; integer ambiguity resolution; multi-constellation; sensor integration; tightly-coupled navigation system.