The invention provides an uncertainty robot adaptive neural network control method based on an inversion method. The method includes the steps of step 1, establishing a kinetic model of an uncertaintyrobot; step 2, converting the kinetic model of the uncertain robot into a general state space form; step 3, defining a tracking error, and converting the general state space form in the step 2 into astate space form about the error; step 4, designing a model-based controller by using the inversion method; step 5, introducing a Moore-Penrose inversion, and designing a control law tau 0; and step6, designing an adaptive controller by utilizing a neural network under the condition of considering the kinetic model and the uncertainty of external disturbance. According to a neural network adaptive trajectory tracking method and the controller provided, the disturbance generated by an external environment and robot modeling uncertainty are fully considered, so that the method and the controller have more practical significance for improving the adaptive ability of a robot control system to uncertainty factors.