3D dynamics analysis of a golf full swing by fusing inertial sensor and vision data

In this paper, a golf swing motion tracking system is proposed, where golf club trajectory (position and velocity) and club face orientation information are given. Infrared LEDs and an inertial sensor unit are attached on a golf club and a stereo camera unit is used. During the address and impact golf swing phases, the cameras capture infrared LEDs on a golf club. Using infrared LEDs as landmarks, position and orientation of a golf club is computed. When a golf club is moving, an inertial navigation algorithm is used to compute the golf club trajectory. An indirect Kalman filter with 9 states is used to combine inertial sensor and vision data. The average position accuracy is about 3.6cm and the maximum error is about 13.2cm. The proposed system can be used to analyze the golf swing quantitatively.