A robot must be better than a human driver: an application of fuzzy intervals

Fuzzy control is usually based on the experience of a human operator. This experience is, however, limited, because a human controller can control only a few parameters at a time. A computerized automated system can potentially control as many parameters as necessary. In particular, when controlling a moving robot, it is potentially possible to control each wheel separately, and thus obtain its maximum mobility. There are many applications in which the moving robots must be manoeuvred in tight spaces where no space is available for turns. If one tries to apply fuzzy control methodology to this complicated motion, then, due to the fact that there is not much experts experience, one gets very crude control rules. As a result, lots of wiggling and wobbling motions are produced by the crude control strategy. In this paper, a simple adaptive fuzzy logic controller is presented to avoid the wiggling. The adaptive fuzzy logic controller takes into consideration the fact that the degrees of belief can be determined only approximately, and form the resulting interval (u^- ,u⁺) of possible control values. A value u that leads, e.g., to the smallest energy consumption is selected as the control. As part of the demonstration, a video of an actual fuzzy-controlled mobile robot will be presented