End-to-End Autonomous Exploration with Deep Reinforcement Learning and Intrinsic Motivation

Table 4

The secondary training results for TC-network.

Method	Environment	TC-network (%)
Pretraining	Parameter selection	92.36
	Maze-1	84.52
	Maze-2	85.14
	Maze-3	78.32
Targeted training	Maze-1	93.16
	Maze-2	92.67
	Maze-3	92.03
Generalization training	Maze-1/Maze-2	90.89
	Maze-1/Maze-3	91.35
	Maze-2/Maze-3	90.62
	Maze-1/Maze-2/Maze-3	90.28