Fuzzy interval control of mobile robots

A new fuzzy logic control (FLC) methodology called the fuzzy interval control (FIC) is discussed in this paper. The FIC implements an interval-valued FLC (or type 2 FLC) without the need to evaluate the interval-valued fuzzy sets. The structure of the FIC consists of a conventional FLC (or type 1 FLC) operating in the normalized universe of discourse; a set of parameters called the sensitivity indices which determine the intervals of the membership functions in the normalized universe; a set of input and output fuzzy gains which control the mapping of the normalized universe to the real axis; and a performance optimizer which dynamically adjusts the values of the sensitivity indices and input and output fuzzy gains in the run time. The invariant properties of the interval mapping preserve the stability of the underlying FLC in the FIC. The significance of FIC methodology is threefold. First, the FIC represents an adaptive FLC whose input-output relation is no longer deterministic. The adaptability of the controller allows the underlying FLC to be extremely simple and fast. This characteristic enables the FIC to be implemented on low-cost embedded microcontrollers for cost-sensitive industrial applications. Second, the FIC is designed for the applications in which limited or uncertain expert's experience is available. Finally, the FIC provides a practical implementation of a type 2 FLC to real applications. To validate the FIC methodology, an autonomous navigation system consisting of a fuzzy logic implemented navigator and an FIC autopilot are designed and implemented on Motorola 68HC11 8-bit microcontrollers. This system has successfully navigated a miniature robot in an unknown maze without touching the walls. Copyright (C) 1996 Elsevier Science Ltd