A Fusion Method of Local Path Planning for Mobile Robots Based on LSTM Neural Network and Reinforcement Learning

Table 4

Orthogonal experimental design and results.

No.	LR	BS	TT	HL	Training set			Test set			Success rate (%)
					R^2	RMSE	Loss	R^2	RMSE	Loss
1	0.1	10	200	L128	0.6246	0.0944	0.0089	0.66	0.0941	0.0089	31
2	0.1	20	300	L150	0.1955	0.1383	0.0191	0.198	0.1445	0.0209	25
3	0.1	50	400	L100 F50	0.0447	0.1507	0.0227	0.0739	0.1553	0.0241	21
4	0.1	100	500	L128 F64	−0.2102	0.1696	0.0288	−0.2188	0.1782	0.0317	9
5	0.1	150	600	L100 L50	0.8379	0.0621	0.0039	0.8082	0.0707	0.005	42
6	0.05	10	300	L100 F50	0.5611	0.1021	0.0104	0.5769	0.105	0.011	33
7	0.05	20	400	L128 F64	0.7094	0.0831	0.0069	0.7613	0.0789	0.0062	35
8	0.05	50	500	L100 L50	0.8929	0.0504	0.0025	0.84	0.0646	0.0042	47
9	0.05	100	600	L128	0.907	0.047	0.0022	0.8528	0.0619	0.0038	60
10	0.05	150	200	L150	0.7686	0.0742	0.0055	0.7701	0.0774	0.006	28
11	0.01	10	400	L100 L50	0.9327	0.04	0.0016	0.8741	0.0573	0.0033	75
12	0.01	20	500	L128	0.9177	0.0442	0.002	0.8757	0.0569	0.0032	63
13	0.01	50	600	L150	0.9381	0.0384	0.0015	0.8841	0.055	0.003	72
14	0.01	100	200	L100 F50	0.9128	0.0455	0.0021	0.8714	0.0579	0.0033	75
15	0.01	150	300	L128 F64	0.9359	0.039	0.0015	0.8748	0.0571	0.0033	92
16	0.005	10	500	L150	0.9351	0.0393	0.0015	0.8918	0.0531	0.0028	80
17	0.005	20	600	L100 F50	0.9336	0.0397	0.0016	0.8836	0.0551	0.003	91
18	0.005	50	200	L128 F64	0.8797	0.0535	0.0029	0.8629	0.0598	0.0036	76
19	0.005	100	300	L100 L50	0.945	0.0362	0.0013	0.8875	0.0541	0.0029	86
20	0.005	150	400	L128	0.9463	0.0357	0.0013	0.8778	0.0564	0.0032	97
21	0.001	10	600	L128 F64	0.9593	0.0311	9.67E-04	0.8887	0.0538	0.0029	88
22	0.001	20	200	L100 L50	0.9213	0.0432	0.0019	0.8676	0.0587	0.0034	78
23	0.001	50	300	L128	0.9107	0.0461	0.0021	0.8703	0.0581	0.0034	91
24	0.001	100	400	L150	0.9053	0.0474	0.0022	0.8428	0.064	0.0041	82
25	0.001	150	500	L100 F50	0.8798	0.0534	0.0029	0.8481	0.0629	0.004	74