Comparative analysis of Type-1 and Type-2 fuzzy control in context of learning behaviors for mobile robotics

Dynamic uncertainties, manifested as input noise or variable environment conditions, are an inherent part of most real world control applications. Recently, several researchers demonstrated that Type-2 Fuzzy Logic Controllers (T2 FLC) are able to cope with such uncertainty and reduce its negative effects. However, the design and optimization of T2 FLC and its subsequent unbiased comparison to T1 FLC are still an open question. This paper presents a comparative analysis of interval T2 (IT2) and T1 FLCs in the context of learning behaviors for mobile robotics. First, a T1 FLC is optimized using the Particle Swarm Optimization algorithm to mimic a wall-following behavior performed by an operator. Next, an IT2 FLC is constructed by symmetrically blurring the fuzzy sets of the original T1 FLC. The performance of the fuzzy controllers is compared using a wall-following sonar-equipped mobile robot in both noise-free and noisy environments. It is experimentally demonstrated that the IT2 FLC can cope better with dynamic uncertainties in the sensory inputs due to the softening and smoothing of the output control surface by the IT2 fuzzy sets. However, the IT2 FLC is outperformed by the T1 FLC when sudden and fast response of the controller is required, such as in the case of turning around corners. Those results suggest the difficulties of symmetrical blurring of T1 FLC (although commonly used) as a design methodology for obtaining the architecture of an IT2 FLC.