建立微巨觀混成模型以了解自駕車決策對車流的影響

A hybrid micro-macro model for understanding the impact of autonomous vehicle strategies on traffic flow

楊雅媛(Ya-Yuan Yang)

指導教授 ： 詹魁元

國立臺灣大學/工學院/機械工程學研究所/碩士(2021年)

https://doi.org/10.6342/NTU202102645

摘要

自動駕駛車輛是未來交通的趨勢，在道路上的佔有比例將會逐漸增加。然而現今沒有針對車輛「須達到什麼樣的門檻才能上路」或是「制定等級以區分上路行駛範圍」的共識與規範，尤其市區的道路設計與車輛互動行為較為複雜，因此我們透過了解車流變化，給予該制度之參考依據。本研究結合微觀車流模型與巨觀車流模型的優點，建立適用於城市的車流混成模型。在此模型中，交叉路口為道路系統之連結點，車流於該處轉向，引發各種交通衝突，故路口以微觀顯示，其餘交通現象則以巨觀模擬。本模型亦同時考量人為駕駛與自動駕駛，以全速差模型表示一般人為操作車輛，用智慧駕駛模型表示自動駕駛車輛。我們經由隊列穩定性、衝擊波、衝突量，分別探討不同決策行為之自動駕駛車輛對車流的影響。模擬結果顯示若自動駕駛車輛的策略過於保守會使車流隊列拉長且難以消散，過於積極則事故風險提高。若車輛間的決策邏輯相差甚遠，則在車流交會點發生事故的可能性也會提升。本論文以一個真實路口的案例模擬，可以得到保守策略自駕車不適合行駛於所有高流量路段，而積極自駕車不適合行駛於高流量又多路口的路段之結論，為自駕車策略的允許駕駛區域提供一個指引。

關鍵字

自動駕駛車輛 ； 決策行為 ； 車流混成模型 ； 交通衝突 ； 隊列穩定性 ； 衝擊波

並列摘要

Autonomous vehicles might ultimately be the mainstream for on-road transportation. However, there are no standards as “what are the technical thresholds to allow autonomous vehicles on public roads” or “how to issue a license to regulate autonomous driving areas”. Moreover, the complex vehicle interactions with the increase of autonomous vehicle penetration urges such rules to be available as soon as possible. Therefore, we develope a hybrid micro-macro model for urban traffic. Under the premise of the complexity of the traffic conflicts, the vicinity of the intersection is on a micro scale, with the rest on a macro scale. All vehicles are considered in the model with regular vehicles use the full velocity difference model, and autonomous vehicles use the intelligent driver model. By observing the string stability, the shock wave and the conflicts, respectively, to discuss the impact of different decision-making behaviors of autonomous vehicles on traffic flow. Our simulation results show that an overly conservative autonomous driving strategy will increase the queue, while an overly aggressive strategy will increase the risk of accidents. In addition, if the decision logics between vehicles are very different, the possibility of accidents will also increase. Combined with the case simulation of real intersection, conservative decisions are not suitable for high-volume roads, while aggressive ones are not suitable for high-volume roads with multiple intersections. The results of the study can provide a guide to the permitted driving area for autonomous vehicle strategies.

並列關鍵字

Autonomous vehicle ； Decision-making behavior ； Hybrid micro-macro model ； Traffic conflict ； String stability ； Shock wave

參考文獻

[1] SAE international, “SAE J3016 automated-driving graphic.” https://www.sae.org/news/2019/01/sae-updates-j3016-automated-driving-graphic, 2019. [Online; accessed 28July2020].

[2] R. Hussain and S. Zeadally, “Autonomous cars: Research results, issues, and future challenges,” IEEE Communications Surveys Tutorials, vol. 21, no. 2, pp. 1275–1313, 2018.

[3] M. Bertoncello and D. Wee, “Ten ways autonomous driving could redefine the automotive world,” McKinsey Company, vol. 6, 2015.

[4] T. Litman, “Autonomous vehicle implementation predictions: Implications for transport planning,” Victoria Transport Policy Institute Victoria, Canada, 2021.

[5] C.Y. Chan, “Advancements, prospects, and impacts of automated driving systems,” International journal of transportation science and technology, vol. 6, no. 3, pp. 208–216, 2017.