ĐIỀU KHIỂN ROBOT VỚI MẠNG HÀM BÁN KÍNH CƠ SỞ CÓ HỆ SỐ HỌC ĐƯỢC TỐI ƯU BẰNG THUẬT TOÁN DƠI - BAT

Abstract

This paper investigates the use of the Bat Algorithm (BAT) to optimize control parameters in a robot controller utilizing Radial Basis Function Networks (RBFN). Previous studies often selected learning coefficients based on experience, which typically required significant time to find coefficients that meet the control problem's requirements and did not guarantee optimality. The Bat Algorithm (BAT), inspired by bats' use of echolocation for navigation, has been applied to globally optimize the learning coefficients for the RBFN. Simulation results show significant improvements in convergence speed and control quality compared to traditional methods.