Adaptive One-Step Prediction Enhanced Kalman Filter Algorithm for INS/Vision Integrated Mobile Robot Localization

Localization accuracy is an important index for evaluating multi-sensor integrated system. For ensuring the high precision, usually more accuracy sensor is adopted for reducing the measurement error or better localization algorithms are used. This paper proposes an adaptive one-step prediction enhanced algorithm that combines kalman filter (KF) algorithm for INS/Computer Vision integrated localization system, to address the issue of some data in the inertial navigation system (INS) being unusable for data fusion. The process can be realized by determining whether one-step prediction is needed by the fast or slow motion of the robot. Based on that, the optimal estimation of the system state can be achieved by computer programming. Analysis of computer simulation experimental is done by comparing the proposed algorithm with INS/Computer Vision integrated KF and one-step prediction enhanced KF, the positioning accuracy can be enhanced and the positioning error can be minimized.