The invention discloses an indoor positioning method based on improved square root unscented Kalman filtering. The method comprises the following steps of firstly, establishing a UWB positioning modeland a pedestrian dead reckoning PDR model in a fusion indoor positioning method; then, according to a residual identification method criterion, determining whether an NLOS criterion exists in a positioning result of the UWB positioning model; if the NLOS exists, using an improved square root unscented Kalman filtering algorithm to fuse a coordinate of inertial navigation dead reckoning with a positioning coordinate of UWB, and correcting a UWB solving coordinate data; and if the NLOS does not exist, using the UWB to analyze a current coordinate and completing real-time positioning. In the invention, the pedestrian dead reckoning PDR is used to compensate a UWB signal loss or non-line of sight condition, which improves the positioning accuracy of the system and the advantages of high precision and high robustness are possessed.