The cooperative positioning method based on wireless sensors has become a popular positioning solution for the intelligent vehicle in recent years, especially in GNSS-denied environments. However, the error due to the Non-line-of-sight (NLOS) propagation is the main problem which affects cooperative positioning accuracy noticeably, thereby affecting the intelligent vehicle’s functions. In order to circumvent aforementioned problem, this paper proposes a multi-sensor cooperative fusion positioning methodology to achieve reliable position in NLOS environments. Initially, the NLOS identification module based on the consistency of multi-sensor measurement value is used to detect NLOS signals. Then, the NLOS mitigation module based on multilayer perceptron is implemented to inhibit the NLOS error by decoupling the relationship between the motion of intelligent vehicle and the change of UWB observations. Lastly, the adaptive fuzzy factor graph fusion module is designed for multi-sensor fusion to overcome the influence of variable signal quality. The real vehicle experiments have been conducted. Compared with the traditional least squares method, the Root Mean Square Error (RMSE) is reduced from 5.932m to 1.056m, and the Maximal Error (MAX) is reduced from 38.886m to 5.583m. The results have proven the effectiveness of our proposed methodology.