摘要：AbstractThe automation of ships offers great potential in terms of safety and efficiency. A river ferry is automated for this reason in the research project AKOON. The performance of the ferry’s guidance, navigation and control (GNC) system highly depends on state estimates which act as a feedback signal and close the control loop. To estimate the ferry’s state in an accurate, precise and robust fashion, sophisticated navigation techniques are required. Therefore, this paper addresses the state estimation of the river ferry by providing a new navigation filter concept. The filter fuses quantities of an inertial measurement unit (IMU) and pseudo- and deltaranges of a dual antenna GNSS receiver. The estimation of heading of the ferry with slow dynamics poses a challenge due to bad observability. The filter is extended by a new correction step which integrates the heading estimates of a GNSS compass in a loosely-coupled manner to overcome this problem. The filter is implemented as unscented Kalman filter (UKF). For automatized operation, the UKF must achieve a maximum horizontal position accuracy of 1 m and a maximum heading accuracy of 1 deg in the river crossing and approach phase. Hence, the concept is evaluated with measurement data from real world tests with the river ferry during daily operation. Furthermore, the experimental setup is discussed in detail. During multiple river crossings, a maximum 2D position error of 52 cm and a maximum heading error of less than 1 deg are achieved.
