Adaptive Inverse Optimal Control for Rehabilitation Robot Systems Using Actor-Critic Algorithm

Table 2

Summarized experimental results of five volunteers.

Sub.	Maximum				RMS error
	NN + AC		NN + AHC		NN + AC		NN + AHC
	PEr		PEr		PEr		PEr
Units	cm	N	cm	N	cm	N	cm	N
A (L)	2.12	9.96	2.02	9.33	0.67	0.26	0.66	0.26
A (R)	2.16	10.15	2.10	9.82	0.68	0.27	0.65	0.27
B (L)	1.92	7.08	1.71	6.36	0.75	0.28	0.56	0.19
B (R)	1.96	7.15	1.73	6.53	0.77	0.28	0.48	0.16
C (L)	2.33	9.13	2.23	8.70	0.86	0.34	0.73	0.27
C (R)	2.31	9.03	2.27	8.64	0.78	0.31	0.64	0.25
D (L)	1.78	4.64	1.73	4.53	0.42	0.11	0.56	0.16
D (R)	1.76	4.59	1.72	4.57	0.39	0.10	0.41	0.11
E (L)	2.02	5.83	2.02	5.62	0.62	0.23	0.56	0.17
E (R)	2.03	5.86	1.96	5.65	0.58	0.21	0.58	0.18
Mean	2.04	7.34	1.95	6.98	0.65	0.24	0.58	0.20
S.D.	0.20	2.34	0.16	1.98	0.12	0.07	0.09	0.05

Note: A-B denotes five volunteers, L represents left arm, R represents right arm, PEr represents position tracking error, and represents the result force; S.D. denotes statistically significant value.