The invention provides a potential-energy-field-based robot path tracking algorithm. The algorithm comprises: (1), a turning radius is obtained; (2), on the basis of the turning radius R and a current linear velocity Vx(t), a linear velocity Vx(t+1) at a next time is estimated; and (3), according to the R and the Vx(t+1), an angular velocity omega is calculated, a reasonable path tracking velocity (Vx, omega) is obtained, and then fitting is carried out on an obtained current path and a preset path to obtain an optimal path tracking velocity (Vx, omega optimal), thereby determining an optimal tracking path track. According to the invention, the basic velocity decomposition value (Vx, omega) is obtained based on analysis and calculation on the real-virtual path of the robot; and then the optimal path is predicted by the robot path in the potential energy field, so that the calculation efficiency is improved substantially and the requirement on the computer hardware is reduced. Moreover, no matter how the characteristics of the system or how large the initial deviation angel of the robot relative o the preset path is, convergence of the robot to the preset path can be guaranteed by the algorithm.