Distributed Strapdown Inertial Navigation System Error and Inertial Sensory Bias Compensation

In this paper, error dynamic model of Strapdown inertial navigation system (SINS) is employed for error compensation of Strapdown algorithm. Perfect visual sensor data is fused with inertial sensors to produce deviation vectors as noisy measurement models. Due to the high dimensional and sparse error dynamic, the system is decomposed to cascaded subsystems because of the system structure. Then, distributed (cascaded) Kalman filters (KFs) and state feedback compensators are designed according to interactions of subsystems. This not only speeds up computations and avoids error propagation but also makes tuning, debugging, and the verifying of the algorithm from the perspective of implementation easier and more precise. The proposed architecture is appropriate to be implemented by multiple processors. The experimental results based on data from 3D MEMS IMU and camera system are provided to demonstrate efficiency of the proposed method.