Adaptive Traffic Signal Control Model on Intersections Based on Deep Reinforcement Learning

Table 6

Improvements in measures of our model relative to those of other methods for multi-intersection case.

SL no.	Method	Reward (%)	Average travel time (%)	Average speed (%)
1	DQN (base)	5.0	−10.0	12.7
	Q-learning	70.5	−41.7	94.5
	LQF	7.2	−11.8	29.0
	Webster	23.9	−17.0	32.4
2	DQN (base)	9.2	−16.2	19.0
	Q-learning	63.1	−40.4	86.2
	LQF	>100	−48.8	>100
	Webster	60.2	−35.1	63.7
3	DQN (base)	24.1	−31.8	44.3
	Q-learning	62.6	−42.6	83.3
	LQF	>100	−52.8	>100
	Webster	58.9	−33.5	54.4