Multi-Inertial Sensor-Based Arm 3D Motion Tracking Using Elman Neural Network

Table 1

Performance indices of each test set.

	-axis			-axis
	Wrist
FSL	10.7	35.1	0.90	52.1	203.5	0.90
LSL	15.4	41.2	0.85	58.7	196.8	0.85
FS	27.2	129.9	0.94	10.9	24.7	0.85
ESL	28.2	58.5	0.90	26.0	160.6	0.92
Elbow
FSL	11.1	32.0	0.70	36.5	90.3	0.88
LSL	5.6	14.3	0.88	23.3	51.3	0.91
FS	30.1	83.4	0.89	7.5	25.5	0.95
ESL	21.0	53.3	0.93	9.3	22.4	0.92
Shoulder
FSL	12.8	32.5	0.91	13.3	34.2	0.94
LSL	5.0	21.2	0.69	5.7	16.8	0.69
FS	22.0	59.3	0.94	16.9	46.8	0.68
ESL	18.6	49.2	0.75	9.4	28.1	0.90

The data of each motion is separated into train set (70%) and test set (30%), and the performances are calculated based on each test set. The italicized values highlight the models that perform worse.