INS/RTK Outdoor Localization Based On Adaptive Predictive Enhanced Kalman Filter

Localization accuracy is one of the key indicators for evaluating the performance of multi-sensor integration systems. To ensure high accuracy, two strategies are typically employed: first, selecting higher-precision sensors to reduce measurement errors; second, using superior localization algorithms to optimize data fusion and error compensation. With the continuous development of navigation technologies, Inertial Navigation System (INS) and Real-Time Kinematic (RTK) positioning have become commonly used outdoor localization solutions for mobile robots and automation systems. Due to the accumulation of errors over time in INS, while RTK can provide high-precision positioning information, the integration of both can effectively compensate for their respective shortcomings. This paper proposes an Adaptive prediction enhanced Kalman Filter (KF) algorithm based on the fusion of INS and RTK data. By combining the positioning information provided by INS and RTK, the proposed method improves the localization accuracy of AGVs. In the experiments, the superiority of this method in improving localization accuracy is validated by comparing it with the INS/RTK Integrated KF algorithm.