Design and application of an analog fuzzy logic controller

teristics of the active devices in analog circuit, the fuzzy elements can be implemented in very simple structures. This brings a reduction in the circuit complexity which implies better speed performance and reduced chip area consumption. Until now the main drawback of the analog approaches has been their poor flexibility. Some programmable analog fuzzy chips have been suggested. Yamakawa et al. programs the membership functions and the fuzzy rules over digital registers which needs a rather large chip area. With a 2μm BiCMOS technology, the die size of the fuzzy inference part (rule chip) is 11.0 x 10.9 mm2 while the defuzzifier chip takes an area of 3.4 x 2.9 mm2 with a 3 μm bipolar technology. The processing speed of the fuzzy system is 1,400,000 fuzzy inference/s without defuzzification and 625,000 fuzzy inference/s including defuzzification [9]. Kettner et al. generates the layout from existing generic macrocells which are programmable only during the prototyping phase [6]. Ungering et al. introduce a mixed-mode fuzzy controller with a digital inference unit to achieve more flexibility. To avoid A/D and D/A converters, they used sorted pointers instead of the analog membership values. Unfortunately, this limits the number of the overlapping membership functions. By using FPGAs (XILINX 4010) for the digital part and Opamps for the analog part, an operation speed of some 10KHz was achieved [23]. To overcome these drawbacks, we propose a novel architecture for the hardware implementation of fuzzy logic control. After a short summary of the fuzzy inference algorithm, which forms the theoretical background of our hardware, we give a description to the new analog approach in section III. To get a better understanding of the advantages of the architecture presented, a real-time collision avoidance for an autonomous system is chosen. In section IV a brief description to the autonomous system is given. For this kind of application not only the processing speed is of great importance but also the flexibility of the implemented rule-base. With changing goals and/or environment new rule-bases are required. The system implementations of the control strategy in analog and digital techniques are presented in section V. A comparison between the two approaches concludes the paper.