提出了一种新颖的鲁棒自适应控制策略，采用简单的多项式结构，不需传统的回归矩阵计算，而唯一需要了解的只是系统的阶数和输出的位置及速度状态，能够有效地克服通常难于建模的摩擦力和外部扰动影响，最后保证全局的指数稳定或全局一致最后有界．二自由度机器人的仿真证明了该法的有效性． A novel robust adaptive control strategy is proposed, which adopts simple polynomial structure without traditional regression matrix calculation. The only thing that needs to be understood is the system order and output position and velocity state, which can effectively overcome the usual Difficult to model the friction and external disturbances, and finally ensure that the global exponential stability or global consensus finally bounded. The simulation of a two-degree-of-freedom robot proves the validity of this method.