This paper proposes a positioning mode switching algorithm for positioning tags based on the joint system confidence and motion trend. The algorithm is based on establishing the position estimation confidence of the UWB/GNSS system and the confidence fusion seamless positioning algorithm of particle filtering, which uses a combination of hard and soft switching to achieve seamless positioning. Through the experiment, the positioning accuracy of the positioning tag is measured to be 50.6cm and 56.4cm in the indoor area and the stationary state in the mixed area, with the standard deviation less than 32cm and less than 36cm, respectively; the walking trajectory is stable and clear in the indoor and outdoor dynamic test, and the positioning switching effect is obvious.