In order to improve the control accuracy of the contact force when the manipulator interacts with the environment, a sensorless active compliance control method based on expected trajectory correction was proposed. Aiming at the problems of traditional parameter identification methods, such as many saddle points of loss surface function, easy to fall into local optimization, and large calculation amount, based on Newton-Euler method, a recursive parameter identification method of manipulator from outside to inside was proposed, which could realize the step-by-step and batch identification of parameters. Aiming at the active compliance control problem when the manipulator interacts with the environment, the position impedance control principle was used to modify the desired trajectory, and the nonsingular terminal sliding mode control was designed under the condition of parameter uncertainty. Finally, the stability of the control system was proved based on Lyapunov theory. Through experimental verification, the cumulative absolute error of the back calculation torque based on the traditional identification method is 9.35 N·m, while it is 1.05 N·m based on the recursive identification method, which indicates that the accuracy of the parameter recursive identification is much higher than that of the traditional identification method. Under the control of position impedance , the contact force reaches the set value in 2.8 s, and the error fluctuation range is ±0.4 N.The experimental results verify the superiority of the active compliance control method.