Abstract: In order to enhance the integrated performance of multi-sensor multi-target tracking system, it is extreme important to study on the theory and method of spatial alignment. For tracking and location system in short range, it is very difficult for spatial alignment of multi-sensor because of the fast velocity, short data and short track. It often happens that the measurement data has been over before the convergence of the spatial alignment process. This paper proposes a new method, in which the track iteration is employed to solve the difficulty brought by short track. The proposed method links a few short tracks together, and keeps system biases changeless for several points. Based on the above, the proposed method improves the UKF(Unscented Kalman Filter) algorithm properly to complete the spatial alignment algorithm in the short-range system formed by radar and infrared sensors. In computer simulation, four tracks are set up according to proper curve moving model, which is used to validate the efficiency of the proposed method and the convergent performance of system biase. To sum it up, the performance of the improved UKF, TI_UKF is better than the improved EKF.