Research on the Agricultural Machinery Path Tracking Method Based on Deep Reinforcement Learning

Table 1

Experimental parameter setting table of the mobile agricultural machinery robot.

Parameter name	Parameter assignment
Reward discount rate	0.9
Actor network learning rate	0.001
Critic network learning rate	0.001
Priority parameter	0.6
Correction error parameter	0.4
Target network delay update TAU	0.001
OU explores noise
Experience playback pool capacity	200000
BATCH_SIZE	256
Optimizer	Adam
Maximum travel distance per round	3m
Total rounds	15000
Experience pool capacity	50000
Batch capacity	32