Design of composite adaptive controller with multilateral adaptive learning mechanism

To enhance trajectory tracking performance for affine nonlinear systems with parametric uncertainties and improve parameter convergence under interval excitation, this paper proposes a multilateral cooperative adaptive learning mechanism. The initial parameter values are assigned based on available data distribution or predefined bounds when unknown. A composite learning adaptive controller estimates system uncertainties using multilateral learning outputs. Adaptive update laws for unknown parameters and multilateral weights are designed using parameter estimation errors and approximation errors, with a saturation function constraining weight variation rates to suppress oscillations. Experimental results on an inverted pendulum system demonstrate the superiority of the proposed controller over two conventional adaptive controllers.