The invention discloses a non-singular terminal sliding mode force position control method for a constraint-oriented reconfigurable manipulator, and belongs to the field of robot control methods and constrained system control methods. In order to solve the problems of low tracking precision, low convergence speed and buffeting in a conventional terminal sliding mode control force position control method, on the basis of building a system dynamic model of the constraint-oriented reconfigurable manipulator, a novel non-singular terminal sliding mode function is proposed, an RBF neural network is introduced for compensating unknown nonlinear items of a system, coupling items between joints and uncertain items of the model, and the non-singular terminal sliding mode force position control method is invented, so that the trajectory tracking error is converged to zero in finite time; and the controller has relatively good robustness, so that the buffeting effect of the controller is effectively suppressed, a control signal becomes smooth in the whole process, the trajectory tracking precision is ensured, and high-precision and micro-buffeting force position control of the reconfigurable manipulator system is realized.