When a robotic arm performs a mixed tracking task of position/force， position control is generally used for the free motion stage of the robotic arm， while force control is generally used for the constrained motion stage. This control structure of position/force switching can achieve the precise control of the robotic arm's position before contact with the environment， and ensure the accurate tracking of the desired control force after contact. However， due to the switching instability of the switching system itself， the robotic arm's actuators may vibrate or even bounce on the surface of the environment when the robotic arm contacts with the environment at a certain speed. To address this issue， a semi-active damping impedance learning method is proposed， which mainly consists of two parts： a semi-active damping controller based on the position/force switching control； an impedance learning algorithm based on a broyden-fletcher-goldfarb-shanno （BFGS） method， which adjusts the semi-active damping according to the learned environmental parameters to suppress vibration and ensure a smooth transition of the robotic arm on the contact surface. In simulations and experiments， the proposed method is applied to enable the robotic arm to interact with different environments， and the results show that the method can effectively suppress overshooting force during the contact transition stage， prevent vibration during the switching process， and achieve smooth contact and transition.