Map-Based Visual-Inertial Localization: Consistency and Complexity

Drift-free localization is essential for autonomous vehicles. In this letter, we address the problem by proposing a filter-based framework, which integrates the visual-inertial odometry and the measurements from the pre-built map. In this framework, the transformation between the odometry frame and the pre-built map frame is augmented into the system state vector and estimated on the fly. Besides, we maintain the map keyframe poses and employ the Schmidt extended Kalman filter to update the state partially so that the uncertainty of the map information can be consistently considered with low computational complexity. Moreover, we theoretically demonstrate that the ever-changing linearization points of the estimated augmented state make the original four-dimensional unobservable subspace vanish, leading to the inconsistent estimation in practice. To relieve this problem, we employ the first-estimate Jacobian (FEJ) technique to maintain the correct observability properties of the augmented system. Furthermore, we introduce an observability-constrained updating method to compensate for the significant accumulated error after the long-term absence of map-based measurements. Finally, by evaluating the system through both simulation and real-world experiments, we confirm that the system has good consistency and low computational complexity.