In autonomous applications, global navigation satellite systems (GNSS) aided inertial navigation system (INS) utilizing an extended Kalman filter (EKF) is the most widely investigated solution for high-rate and high-accurate vehicle states estimation. However, such navigation system suffers from poor parameterization, environment disturbances, human error, or even software and hardware failures under worstcase scenarios. In this paper, a novel scheme of multi-sensor navigation system is proposed, contributing to following research questions: 1) How to provide a reliable state estimation under minor system aberrations, i.e. improve the robustness of navigation system against e.g. inappropriate parameterization or environment disturbances; 2) How to provide system integrity against worst-case scenarios, i.e. significant system aberrations or even failures. The proposed scheme involves extended H∞ filter (EHF) for robustness enhancement, zonotope for protection level (PL) generation of the navigation solution and vehicle dynamic model aided fault detection (FD) of the inertial sensor. The designed approach is validated using the recorded data from an experimental platform called 'IRT-Buggy', which is an electrical land vehicle. The results show that the proposed scheme provides reliable integrity monitoring and accurate state estimation, under both real-world and artificial abnormalities and shows significant advantages against conventional 'GNSS+INS+EKF' approach.