Multirobot Coverage Path Planning Based on Deep Q-Network in Unknown Environment

Table 1

The average results of the last ten episodes when reward parameters change.

R1 = 1			Average results from the 191^th episode to the 200^th episode
R2	R3	R4	Coverage rate	Repetition rate
10	0	−10	96.8%	378%
10	−0.01	−1	100%	163.4%
10	−0.01	−10	98.8%	306.8%
10	−0.1	−1	100%	170.3%
10	−0.1	−10	100%	197.7%
100	−0.01	−1	97.4%	382%
100	−0.01	−10	95.4%	356.3%
100	−0.1	−1	92%	422.3%