交通資訊與路徑選擇：用路人與管理者之實驗賽局

Traffic Information and Route Choice：Experimental Games between Users and Traffic Management Agent

鄭妍妍(Yen-Yen Cheng)

指導教授 ： 董啟崇

淡江大學/管理學院/運輸管理學系碩士班/碩士(2009年)

https://doi.org/10.6846/TKU.2009.00699

摘要

回顧近年來實際環境中交通資訊的發布結果，發現路網出象往往非管理者與用路人所預期之狀態；因此本研究回溯到路徑資訊的來源(功能與策略)與用路人資訊遵循的路徑決策之競合關係，以賽局理論觀點探討管理者交通資訊的發布與用路人路徑決策的反應。因管理者是基於路網績效目標來發布交通資訊，但管理者該如何發布以達成其預期所想達成之理想狀況，乃有賴於用路人對交通管理資訊之認知與反應決策之集體效應，因而於此情況下衍生出所謂管理者與用路人(集體)間之互動賽局，即為本研究之雙參賽者賽局(Two-Player Game)形式。

本研究之工作內容主要可分為三部份，第一部份分別對交通管理者與用路人進行賽局前置分析調查，主要是為在進行動態模擬實驗前，清楚了解實驗中之代理管理者與實驗者(用路人)之基本行為特性與分別對交通資訊與交通管理策略之認知與態度而進行的前置作業，亦可稱之為控制實驗之先驗知識。第二部份則藉由逐日實驗觀察與紀錄以代理管理者發布的交通資訊與集體用路人之路徑決策結果，了解特定管理策略下用路人路徑決策演變情形。第三部份則藉由實驗所得之決策資料與結果以無意帶區間之概念構建個人路徑轉換行為，進一步了解用路人路徑決策行為是否受管理者發布資訊的策略目標或其他個人變因之影響。

藉由第一部分賽局前置分析可得目前交通單位亦期望以交通資訊作為一交通管理策略，改善路網狀態，且管理者在研擬交通管理策略時，是會考量用路人可能的路徑行為反應進而更動其管理策略；而用路人之路徑決策是會根據前次路網出象而反應，若前次路網出象不符合其預期狀態時，多數用路人是會質疑管理者所發布之交通資訊，進而不遵循管理者所提供之建議路徑；部分用路人即使路網符合其預期狀態，但不符合管理者預期狀態時，亦會對管理者產生質疑。故藉由賽局前置分析可知管理者與用路人彼此間是有賽局存在的。

而藉由第二部份逐日實驗結果可發現若管理者以交通資訊為其交通管理策略時，路網出象會不同於Bi-level之Stackelberg賽局假設，如路網出象未達成管理者預期之系統最佳化狀態；並藉由實驗亦發現於公布交通管理策略意涵時路網出象會較無公布時好；而實驗者於公布交通管理策略意涵之資訊遵循率較未公布管理策略意涵高。
此外，本研究藉由第三部分路徑轉換行為模式之構建，以管理者觀點了解用路人的路徑行為反應會受何種變因(誘因)影響，使其離開習慣路徑。結果顯示用路人於不同管理策略目標與不同資訊完全度時，路徑變換行為會受個人特性影響，包括個人賽局傾向；亦會受其個人累積經驗、近視屬性影響。

關鍵字

實驗賽局 ； 逐日動態路徑選擇行為模式 ； 路徑導引 ； 控制實驗

並列摘要

The dynamic route choice behavior is of the most interest to study the individual driver’s route choices under the influence of the route guidance information where individual driver makes consecutive route switch decisions along the traveling route. This particular issue has been successfully modeled with various forms and extensions under the notion of the “Indifference Bands” applied with Probit model specifications in recent years. However, this particular issue may be further investigated as game-like decisions of two perspectives: the game between road users (among whom makes individual route choice in response to traffic information) as an entity versus traffic management authority (who would distribute traffic and route guidance information to achieve its goal); and the game among road users (individual decision versus predicted others’ actions in response to traffic information).

This study focuses on the game between road users in response to traffic management strategies by authority. The experiment context includes two decision dimensions: day-to-day pre-trip route choice and en-route decisions with a period of ten consecutive working weeks (50 days) under various designated scenarios (sequences). As such, a three stage comprehensive controlled experiment has been conducted where selected subjects were selected to participate in a series of laboratory simulation as commuting drivers while the researcher as acting traffic management authority.

The tasks of this study are in two folds: the first to examine the explicit game where evolution of the individual’s (player’s) consecutive decision and associated playoffs along with aggregate system performance will be recorded and tested including the decision to decision dynamics and their final (convergence) state. The second task is to further modeling the disaggregate road user’s dynamic decision with embedded game concept in the notion of jointed indifference bands (specified as Probit model structure) such that the game elements will be treated as implicit attributes.

The major findings of thesis are summarized in following:

1. The responses from pre-experiment survey confirmed the game-like interactions between traffic management authorities and experiment subjects in terms of general recognitions of intentions and actions of each part.

2. The network reaches approximate user equilibrium UE state after a period of time for providing guidance information based on either UE or SO strategies by the management. However, the result is close to theoretical values with mixed strategy. These results demonstrate that using guidance information as traffic control measures to reach system optimum is unlikely, since it is not as forceful as normal traffic controls. However, a mixed strategy seems more promising.

3. It also shows that while route guidance can be recognized as one of the control strategies by traffic control authorities, its effectiveness is quite different from other strategies in terms of enforcement. As such, bi-level problem formulation with Stackelberg game style accompanying general cases of treating upper level as system optimum for traffic control and lower lever as user equilibrium for route choice is not that promising for the context of this study.

4. Two indifference band models (for partial information and full information) were successfully specified and calibrated with good associated statistical criteria.

並列關鍵字

Day-to-day Dynamic Route Choice ； Experimental Game ； Route Guidance ； Controlled Experiment

參考文獻

蘇秋如，個人化路徑導引資訊：動態路徑選擇行為模式之應用，私立淡江大學運輸科學研究所碩士論文，2006。

鄭珮琪，基於認路原則之車內路徑導引資訊系統之研究，私立淡江大學運輸科學研究所碩士論文，2007。

Adler, J. L., Blue, V. J., “A cooperative multi-agent transportation management and route guidance system,” Transportation Research Part C : Emerging Technologies, Vol. 10, No. 5-6, pp. 433-454, 2002.

Al-Deek, H. M., Khattak, A. J, Thananjeyan, P., “A combined traveler behavior and system performance model with advanced traveler information systems,” Transportation Research Part A : Policy and Practice, Vol. 32, No. 7, pp. 479-493, 1998.

Chen, P. S.-T., Srinivasan, K. K., Mahmassni, H. S., “Effect of information quality on compliance behavior of commuters under real-time traffic information,” Transportation Research Record, Vol. 1676, pp. 53-60, 1999.