Abstract: Aiming at the problem of low efficiency caused by complex and redundant calculation of robot dynamics model, a programmed modeling method(PMM) is proposed. Taking the Stanford Arm with six degrees of freedom as an example, the dynamic model based on Lagrangian equation is established by using this method. According to the core idea of "forward analysis, reverse output", the recursive process of the model is analyzed emphatically. On the basis of verifying the correctness of the model, the indexes such as the dimensions and running time of the Stanford Arm model based on the PMM and the conventional Lagrange equation without the use of the PMM are compared. The results show that relative to the conventional Lagrange method, the complexity of the model established by PMM is reduced by 67.6%, and the computational efficiency is increased by 66.3%. Stanford Arm is a complete constrained system. PMM is extended to underactuated nonholonomic constrained systems, numerical simulation and physical prototype experiment analysis are carried out by using partial feedback linearization control algorithm which is closely related to the model, it′s reliability and effectiveness of the programmed modeling method are verified, which provides a modeling method with higher efficiency and better versatility for different types of robots.