ACTAM: 合作式適性化多代理人系統在都市交通號誌控制之應用

Abstract

當傳統的交通號誌控制系統無法滿足都市的車流狀況後，都市的交通擁塞問題越來越嚴重，所以智慧型運輸系統（ITS）裡所提到的動態交通號誌控制就越來越受到重視。 本研究所提出來的適性化合作式交通號誌控制代理人模型（ACTAM）使用多代理人系統的架構來做分散式交通號誌控制，ACTAM架構是全模組化設計，擁有高彈性和高擴充性的優點，透過模組化的設計，可以視目標設計新的模組加入架構內，或使用更適當的模組取代先前的模組功能。ACTAM的核心為智慧型路口代理人（Intelligent Intersection Agent，IIA），自主式的透過偵測模組蒐集資訊並做處理判斷，為各路口號誌做控制，其內包含了資料層（資料處理模組、溝通模組）、作業層（預測模組、學習模組、權重模組）和決策層（控制策略決策模組）。 我們亦設計一實驗來驗證本研究所提的代理人架構及方法的效能，實驗結果顯示，透過ACTAM方法和架構，使得交通號誌控制新架構擁有主動、即時性調整、動態改變等優點，同時可以減少交通網路內46％的總車輛等待時間。

Urban area traffic congestion problem is getting worse since traditional traffic signal control system could not satisfy the need of efficiently regulate urban area traffic. Therefore, dynamic traffic signal control in ITS have received increasing attention recently. In this paper, we proposed a multi-agent architecture, Adaptive and Cooperative Traffic light Agent Model (ACTAM), for decentralized traffic signal control system. The proposed architecture is composed of data storage and communication layer, traffic regulation factors processing layer, and decision-making layer. In this paper, we focused on through the cooperation and adaptation of multi-agents and the prediction mechanism of our architecture. Through the proposed architecture, dynamic traffic signal control could be achieved. Thus, total delay time of the traffic network under ACTAM, in comparison with conventional fix sequence traffic signal control strategy, can be reduced by 46%. Consequently, traffic congestion problem in urban area can be alleviated through ACTAM.