The class of Pneumatically-driven Low-pressure Soft Actuators (PLSA) is a popular choice potentially used in the surgical robotic applications. One fundamental problem lying in the PLSA research is the lack of a generally validated model for the complex nonlinear dynamic behaviours. In this paper, a unified identification approach for the general PLSAs is proposed. It is a parameter-independent way directly used to identify the dynamical relation between the actuating pressures and the principal degrees of freedom of a PLSA, the bending and the steering. The approach is based on a modified auxiliary kinematic setting and a newly developed identification model structure, named DIO–PWL–OBF. Following the concluded identification procedure, the implementations for the single chamber bending and the double chamber bending and steering are demonstrated separately. The results show that the proposed approach can accurately capture the nonlinear pressure–shape dynamical relation. The approach is also efficient in real-time applications. It can be further used to improve the current control design for the PLSAs in robotic applications.