A GNSS/INS Integrated Positioning Method Based on TCN and Collaborative Adaptive Filtering During GNSS Outages

• 1、School of Computer Science (National Pilot Software Engineering School), Beijing 100876

Abstract：In intelligent transportation applications, high-precision and highly continuous positioning capability in complex environments has become a key requirement for ensuring reliable system operation. To address the rapid error accumulation of GNSS/INS integrated navigation systems caused by GNSS outages in complex scenarios such as urban canyons and tunnels, this paper proposes an integrated navigation method based on a Temporal Convolutional Network (TCN) and cooperative adaptive filtering. Within the framework of the Error-State Kalman Filter (ESKF), a TCN is employed to perform multi-scale temporal modeling of high-frequency IMU data. During GNSS outages, pseudo-observations are constructed to maintain the filter update process. Meanwhile, a cooperative adaptive adjustment mechanism is designed based on the cross-correlation characteristics between the innovation sequence and the state residual sequence, enabling the joint dynamic adjustment of process noise and observation noise covariance matrices, thereby improving the filter\'s adaptability to model mismatch and pseudo-observation uncertainty. Vehicle experiments under GNSS outage durations of 60 s, 90 s, 120 s, and 150 s demonstrate that the proposed method effectively suppresses error growth and significantly reduces the positioning root-mean-square error.

Keywords： GNSS outage GNSS/INS integrated navigation Temporal Convolutional Network (TCN) cooperative adaptive filtering vehicle positioning