In this paper, a new algorithm is proposed for the constrained control of weakly coupled nonlinear systems. The controller design problem is solved by solving Hamilton-Jacobi-Bellman(HJB) equation with modified cost to tackle constraints on the control input and unknown coupling. In the proposed controller design framework, coupling terms have been formulated as model uncertainties. The bounded controller requires the knowledge of the upper bound of the uncertainty. In the proposed algorithm, Neural Network (NN) is used to approximate the solution of HJB equation using least squares method. Necessary theoretical and simulation results are presented to validate proposed algorithm. © 2009 Springer-Verlag Berlin Heidelberg.