At present, Global Navigation Satellite System (GNSS) is the most widely used technology in the field of vehicle navigation and positioning. However, due to satellite signal blocking, it can be challenging to obtain precise and reliable vehicle position in completely GNSS-denied environments. In this paper, a positioning strategy utilizing ultra-wide band (UWB) and low cost MEMS Inertial Navigation System (INS) is proposed, aimed at tracking vehicles in typical GNSS-denied scenarios. UWB tech owns the potential of high-precision ranging through its strong penetrating ability. However, Non-Line-of-Sight (NLOS) propagation still has a high occurrence in typical traffic scenes due to the direct path between a radio transmitter and a radio receiver is obstructed. Therefore, a two-step NLOS error mitigation method is proposed to deal with the abnormal measurements of UWB in tunnels. First, a state-of-the-art probabilistic factor graph model has been proposed to estimate the vehicle’s position information based on UWB range measurements instead of excluding the NLOS receptions. Then, a Residual weighting algorithm (Rwgh) is employed during UWB/INS loosely-coupled integration to further mitigate the NLOS error. Finally, the performance of proposed methodology was evaluated through experiments. Experiments conducted in typical NLOS environment of Nanjing showed that the proposed method could achieve higher positioning accuracy when compared with the conventional UWB/INS integration.