This paper presents a novel spatial registration strategy for multiple motion radars with varying spatial attitudes， which does not require prior information such as radar position and attitude. The strategy involves key steps， including the real-time calculation of registration parameters and optimization of fusion tracks. The recursive least squares （RLS） is employed to establish spatial attitude relationships between radars by using target track data. A method for reverse adjustment of target trajectory-level spatial registration parameters based on fusion results is presented. The method models registration errors and optimizes them using gradient descent， resulting in reduced fusion trajectory errors， improved registration accuracy， and enhanced tracking quality. The proposed strategy provides a comprehensive and efficient solution for real-time registration of radar spatial attitudes. It has both theoretical and practical value.