Design and experiment of a fuzzy PID track-keeping ship autopilot

This paper is concerned with the design and experiment of a fuzzy PID track-keeping autopilot. The proposed approach avoids the time-consuming tuning process for the PID controller gain coefficients and only requires a small number of fuzzy sets (three for the inputs and two for the outputs). Since the gain coefficient of a PID controller can be expressed in terms of the process parameters and a design parameter that characterizes the speed of response of an internal model control (IMC) system. The IMC method is found useful in determining the interval of the universe of discourse of the proposed fuzzy PID controller gains. The line of sight (LOS) guidance law uses the positioning information provided from RTK GPS in computing the reference heading angle needed to guide the ship towards predefined waypoints. The heading deviation and rate of heading deviation are treated as fuzzy inputs and the PID gain coefficients are the fuzzy outputs. Specifically, three sets of input membership functions were used, and both three-set and two-set output membership function design were studied in the four-waypoint track-keeping maneuvers. Simulated results indicate that the achieved track-keeping performance of the two-set output membership functions-based design is nearly as good as that of the three-set output membership function-based design. Similar track-keeping scenarios were tested in the small boat-based track-keeping experiments, which were carried out in the small boat harbor near National Taiwan Ocean University. Experimental results indicate that satisfactory track-keeping performance is achieved for both two-set and three-set designs. This indicates the feasibility and robustness of the proposed approach, moreover the simple structure of the two-set output membership function design makes it attractive for practical implementation.