In response to the outage of Global Navigation Satellite System (GNSS) and the rapid divergence of errors in Inertial Navigation System (INS) in integrated navigation, this paper proposes a localization strategy based on Convolutional Neural Network cascaded Bidirectional Gated Recurrent Unit (CNN-BiGRU) to assist in integrated navigation. When the satellite signal is normal, the INS and GNSS information are used to train the CNN-BiGRU model. When using residual \(\chi^{2}\) detection method to detect satellite signal outage, the trained CNN-BiGRU model is used to output GNSS pseudo position and continuously correct the position error of INS. The results of the sports car experiment show that the adaptability and stability of this algorithm are significantly better than other neural network assisted methods in different road conditions, which fully ensures the positioning robustness of the receiver during GNSS outage.