An innovation gain-adaptive Kalman filter for unmanned vibratory roller positioning

Abstract

The precise positioning of an unmanned vibratory roller is the key to the compaction quality. Currently, many earth-rockfill dams are constructed in deep narrow valleys. Ultra-wide band (UWB) is widely used to overcome the partially denied problem of the global navigation satellite system. However, due to the complex noisy interference on construction sites, the accuracy of UWB is insufficient. To ensure the precise positioning of the unmanned vibratory roller, we propose an innovation gain-adaptive Kalman filter (IG-AKF) algorithm. The IG-AKF can adjust the noise covariance based on the relationship between the historical state statistics and positioning deviation. In addition, an attention decay mechanism that associates more attention to recent states is employed. Moreover, a diversity Harris hawks optimisation is proposed to optimise hyper-parameters in the IG-AKF. The proposed method was applied in Lianghekou project. Compared with UWB and the standard KF, the positioning accuracy of the IG-AKF was significantly improved.