MTLight: Efficient Multi-Task Reinforcement Learning for Traffic Signal Control

Liwen Zhu, Peixi Peng, Zongqing Lu, Yonghong Tian

Keywords: Multi-Agent Reinforcement Learning, Multi-Task Learning, Traffic Signal Control

TL;DR: MTLight is proposed to enhance the agent observation with latent task-specific and task-shared states by constructing multiple auxiliary tasks for traffic signal control application..

Abstract: Traffic signal control has a great impact on alleviating traffic congestion in modern cities. Deep reinforcement learning (RL) has been widely used for this task in recent years, demonstrating promising performance but also facing many challenges such as limited performances and sample inefficiency. To handle these challenges, MTLight is proposed to enhance the agent observation with a latent state, which is learned from numerous traffic indicators. Meanwhile, multiple auxiliary and supervisory tasks are constructed to learn the latent state, and two types of embedding latent features, the task-specific feature and task-shared feature, are used to make the latent state more abundant. Extensive experiments conducted on CityFlow demonstrate that MTLight has leading convergence speed and asymptotic performance. We further simulate under peak-hour pattern in all scenarios with increasing control difficulty and the results indicate that MTLight is highly adaptable.