Multiobjective Evolutionary Learning of Interval Type-2 Fuzzy Controllers for a Wall-Following Wheeled Mobile Robot
Chia‐Feng Juang, Y. Chen · 2023
This paper proposes a multiobjective evolutionary learning approach of interval type-2 fuzzy controllers (IT2FC) to train a wheeled robot to execute the task of wall following. The data-driven approach optimizes a set of non-dominated IT2FCs through the multiobjective non-dominated sorting genetic algorithm II (NSGA II). The employment of the IT2FC instead of a type-1 fuzzy controller is to improve the resistance ability of the controller to sensor measurement noise. In the robot wall-following control task, three objectives are defined, including proper robot-wall distance, high moving speed, and high model interpretability. This paper formulates the learning task as a multiobjective optimization problem and finds solutions through the NSGA II. Simulations are performed to show the effectiveness and advantage of the proposed method.