Improving INS/GPS Integration with Artificial Intelligence during GPS Outage

The importance of navigation precision in high dynamic environments has led to integrating the Inertial Navigation System (INS) with satellite navigation systems. In one of those integration methods that INS is integrated with GPS, GPS outage is an unavoidable challenge. Moreover, due to significant noisy signal existing in low-cost MEMS sensors, navigation precision severely decreases, and the INS error will diverge in the long term. This paper improves the INS/GPS navigation system using Artificial Intelligence (AI) during GPS outage. In this approach, the INS outputs at t and t-1 are injected to the AI module as the positioning and timing information. While GPS is available, the AI module is trained, and its output is compared with the GPS output. The AI module indeed intents to drive the INS output to the GPS output during GPS outage. To evaluate this approach and compare with some different intelligence systems, we have utilized Neural Networks (NNs) as an AI module in five different NNs: multilayer perceptron (MLP(, radial basis function )RBF(, support vector regression (SVR(, Wavelet, and adaptive neuro-fuzzy inference system (ANFIS). The required dataset to compare all five mentioned methods is gathered in a real environment by a mini-airplane. The results of all five methods represent that the proposed methods have superior performance compared to other traditional methods; so that the wavelet NN outperforms others by approximately 30%.