Abstract: A novel trajectory recognition method (TRM) using multi-sensor fusion was proposed to extract the motion trajectory of badminton racket more accurately. The smart device was placed in the bottom of the handle of the racket， which was composed of accelerometer， magnetometer and gyroscope. Firstly， the date of gyroscope was optimized by accelerometer and magnetometer. By combining gyroscope error， the attitude angle of trajectory could be obtained using TRM. Then the speed and the displacement were calculated by frequency domain integral of acceleration data. The accumulated error of frequency domain integral is removed by the least squares fitting method.Finally， the trajectory was identified by attitude angle， the speed and the displacement. Multiple experimental results show that the validity of TRM is great. Compared with the traditional time domain integral trajectory recognition method， the identification accuracy of TRM is higher. TRM can recognize the trajectory of badminton racket more accurately.

Key words: badminton racket, smart device, multi-sensor, attitude algorithm, swing trajectory recognition