A Vertical Channel-Enhanced Fusion Method Based on RINS and Barometric Altimeter for UAVs in GNSS Denial Environments
Xiaoxi Zhao, Yue Meng, Fei Qi, Lei Wang, Xinyao Zhu
Abstract
With the increasing prominence of global navigation satellite system (GNSS) defects in recent years, high-precision inertial navigation systems (INSs) with autonomy are gradually undertaking UAV navigation tasks and are commonly fused with barometric altimeter (BA) to damp the error growth in the vertical channel without GNSS. However, anomalies or failures in barometric altitude lead to the divergence of vertical errors, while indirectly adversely affecting horizontal navigation, bringing new challenges to UAV autonomous navigation. In this article, a vertical channel-enhanced fusion method based on a high-precision rotational INS (RINS) and BA is proposed to enhance the reliability of the RINS vertical channel and correct horizontal navigation information in GNSS denial environments. On the one hand, nonlinear autoregressive with exogenous input (NARX) models for vertical errors are constructed based on Kalman filter (KF) fusion. And the models are identified by least squares support vector machine (LSSVM) with particle swarm optimization (PSO), referred to as NPLSSVM/KF. On the other hand, vertical errors are predicted and compensated by the trained NPLSSVM models during the BA failure, and horizontal navigation errors are also corrected. Finally, the feasibility and superiority of the proposed method are validated by simulation and UAV flight experiments.