TC-DQN+: A Novel Approach to ATSC Using Deep RL
In this repository, we provide the details of the implementation of the following manuscript:
Adaptive Traffic Control with Deep RL: Towards State-of-the-art and Beyond
Siavash Alemzadeh, Ramin Moslemi, Ratnesh Sharma, Mehran Mesbahi
Abstract
In this work, we study adaptive data-guided traffic planning and control using Reinforcement Learning. We shift from the plain use of classic methods towards state-of-the-art in deep RL community. We embed several recent techniques in our algorithm that improve the original DQN for discrete control and discuss the traffic-related interpretations that follow. We propose a novel DQN-based algorithm for Traffic Control (called TC-DQN+) as a tool for fast and more reliable traffic decision-making. We introduce a new form of reward function which is further discussed using illustrative examples with comparisons to traditional traffic control methods.
Case-studies are provided wherein the benefits of our method as well as comparisons with some traditional architectures in ATSC are simulated (from real traffic scenarios).
TC-DQN+ vs Self-Organizing Traffic Light
TC-DQN+ SOTL
TC-DQN+ vs Fixed-Time Traffic Plan
TC-DQN+ FT