Dynamic Programming: Models and Applications
Dynamic Programming: Models and Applications
A Collaborative Reinforcement Learning Approach to Urban Traffic Control Optimization
WI-IAT '08 Proceedings of the 2008 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology - Volume 02
Hi-index | 0.00 |
This paper develops an adaptive, optimal planning algorithm for signal control at a single intersection using an efficient dynamic programming technique. It is called ADPAS (Adaptive Dynamic Programming Algorithm for Signaling). The objective of ADPAS is to minimize the total delay experienced by vehicles passing through an intersection. ADPAS can generate any sequence of green phases to optimize signal control without restriction to fixed cycles of green phases. The algorithm employs reaching as the method to solve the forward DP functional equation, which does not require any prior knowledge of the states of the DP network. The efficiency of the algorithm results from two acceleration techniques that adaptively eliminate inferior states as the algorithm progresses. We verify computational efficiency of ADPAS with several test cases.