Markov Decision Processes: Discrete Stochastic Dynamic Programming
Markov Decision Processes: Discrete Stochastic Dynamic Programming
| Hi-index | 0.00 |
Green-waved traffic control, which coordinates intersections to allow continuous traffic flow from major directions without unnecessary stops, is one of the most efficient strategies to improve the efficiency of urban transportation. However, in large-scale urban transportation systems, traditional centralized coordination methods suffer bottlenecks in both communication and computation. It is hard to make decentralized control with very limited observation to the whole network or avoid conflict traffic flows within a complex intersection. In this paper, we proposed a novel decentralized, multi-agent based schema for massive traffic lights coordination to promote the green-waved effect. In additional, we use constraint optimization agents within a local intersection to efficiently search for joint activities of the lights. This paper also builds a heuristic algorithm to solve the scalable coordination problem and with our massive urban transportation simulator, we have proved the feasibility of our approach.