The Upper-limb Motion Capture Svstem Based On MA And 5DT Glove

The human upper limb motion capture technology is widely used in many fields such as human-computer interaction,disease rehabilitation,motion analysis,brain-computer interface and so on.With the progress of related technology,the stability and accuracy of data have more requirements.In this paper,MA(Motion Analysis)is used to capture the kinematics of the arm.A predict algorithm was adopted to real-time rectify the missing marker points which obtained by MA.This paper uses 5DT data gloves obtain hand kinematics.The arm kinematics and the hand kinematics is integrated to drive a virtual upper limb model.The model real-time reconstruct the human upper limb movement.The specific work of this paper is as following:Calculating the arm joints kinematics by markers position which obtained by the optical motion capture system.The arm was simplified to a 7-DOF model.And the joints positions were calculated using the coordinates of the marked points modles.According to the different situation of markers missed in each frame data,the corresponding prediction algorithm are set up to predict the coordinates of the missing markers in real time.The loss function is established by using the fixed distance between the labeled points on the surface of the skin and the corresponding joints.Least square method is used to solve the rotation center of the marker,which is position of the corresponding joint.The spatial position of each fingers joints are calculated based on the data glove data.In this paper,a 26-DOF hand movement model is adopted to represent hand movements.The rotation angles of each finger DOF are correlated with the value of the data glove sensors.Using the sensor data to calculate the degree of freedom of the finger joint rotation.Finally,the finger joints positions were calculated by the forward kinematics.Unified arm and hand coordinates and real-time drawing upper limb movement.Wrist coordinates are calculated using the marker points of the hand and the forearm.The direction of the hand is calculated using the three marker points of the hand.The motion of the hand is mapped to the arm according to the wrist coordinates and the direction of the hand.Get a complete human upper limb joint movement information.According to this information,using OpenGL to draw the joints and bones.Human upper limb can be synchronized displayed by a virtual upper limb.Through off-line calculation,the loss prediction error markers which obtained by MA basically concentrated within 0.5mm.The joint position marker range at about 1mm.During the long time of 10 minutes,the system can keep stable on-line operation,it is proved that the system can get the real-time data of upper limb.