Pose estimation of an aerial construction robot based on motion and dynamic constraints

High accuracy pose estimation with high data rate of an aerial construction robot is the prerequisite for aerial construction robot control. A dynamic and motion constrained robust extended Kalman filter is developed for robot localization in aerial construction environment which is characterized by radio signal occlusion and few visual features. The motion constraints of the gondola are derived to estimate constrained pose of the gondola while the dynamic constraints of the gondola are introduced to inhibit spike-like data mutation when the gondola shakes back to the measurement range of the self-developed laser spot vision system. Moreover, the robust factor technique is adopted to inhibit instantaneous large outliers. The experimental results show that the pose estimation based on the dynamic and motion constrained robust extended Kalman filter achieves pose measurement of the aerial construction robot with mean position accuracy less than 0.015 m and mean attitude accuracy less than 0.003 rad.