本文選題：自治交叉口　切入點(diǎn)：多Agent系統(tǒng)　出處：《大連理工大學(xué)》2014年碩士論文　論文類型：學(xué)位論文

【摘要】：具備精確控制與傳感能力的自治汽車的出現(xiàn),給安全駕駛帶來了新的希望。當(dāng)前存在的人工智能技術(shù)已經(jīng)能有效的解決自治汽車在開放道路中行駛問題。但面對情景復(fù)雜、擁堵較嚴(yán)重、交通事故多發(fā)的交叉口來說,車輛的安全和交叉口的通行效率目前都無法保證。本文介紹一種自治交叉口管理方案,嘗試?yán)萌蚨ㄎ患夹g(shù)、無線通訊技術(shù)、車內(nèi)傳感和計(jì)算技術(shù)使得自治車輛能夠安全高效地通過交叉口。 控制策略作為整個(gè)自治交叉口管理系統(tǒng)的“大腦”直接影響系統(tǒng)的性能。本文提出了一種基于車路通信預(yù)留的控制策略Batch-Light,該策略能夠充分利用現(xiàn)有的智能交通領(lǐng)域的信號控制技術(shù),自適應(yīng)的應(yīng)對不斷變化的交通流；建立時(shí)空板模型,對車輛的時(shí)空軌跡進(jìn)行數(shù)據(jù)離散化建模；在時(shí)空板模型基礎(chǔ)上,提出一種貪心的沖突矩陣決策算法,在保證不同方向通行優(yōu)先權(quán)的前提下,使得更多的車輛預(yù)留成功。此外,為更大限度程度地提高路口的服務(wù)效率,減少時(shí)間和空間的浪費(fèi),本文又提出一種K平移優(yōu)化算法,來協(xié)助原本無法成功預(yù)留的車輛,盡量避免二次請求,通過加速或減速盡可能的通過交叉口。 本文還對自治交叉口仿真器AIM (Autonomous Intersection Management)進(jìn)行了擴(kuò)展,利用雙向耦合的仿真器互連技術(shù),用NS3進(jìn)行更逼真的網(wǎng)絡(luò)傳輸仿真取代AIM中簡單網(wǎng)絡(luò)傳輸模型,以實(shí)現(xiàn)具備反饋環(huán)的自治交叉口仿真。該工作用以支持研究不同的交通控制策略對數(shù)據(jù)傳輸?shù)男枨蟪潭?以及反過來網(wǎng)絡(luò)傳輸性能對交通控制效率的影響。 最后對本文的提出的控制策略在搭建的仿真平臺(tái)上進(jìn)行仿真驗(yàn)證,實(shí)驗(yàn)結(jié)果顯示Batch-Light無論是在平衡的交通狀況下還是不平衡的交通狀況下都優(yōu)于FCFS (First Come First Served)和傳統(tǒng)的交通信號控制。而且Batch-Light的數(shù)據(jù)傳輸量較小,隨著噪聲的增加,服務(wù)效率雖然有所下降,但性能穩(wěn)定。而FCFS數(shù)據(jù)傳輸量很大,網(wǎng)絡(luò)性能對控制效率的影響不定,有時(shí)隨著丟包率的增加,控制效率反而會(huì)上升。
[Abstract]:The emergence of autonomous vehicles with accurate control and sensing capabilities has brought new hope to safe driving. The current artificial intelligence technology has been able to effectively solve the problem of autonomous vehicles driving on open roads. But the situation is complex. At present, the safety of vehicles and the efficiency of intersections can not be guaranteed. This paper introduces a management scheme of autonomous intersections, which tries to use global positioning technology and wireless communication technology. Internal sensing and computing technology enable autonomous vehicles to pass through intersections safely and efficiently. As the "brain" of the whole autonomous intersection management system, the control strategy directly affects the performance of the system. This paper proposes a control strategy Batch-Lightbased on vehicle-road communication reservation, which can make full use of the existing intelligent transportation. Field signal control technology, Adaptive response to ever-changing traffic flow; establishment of space-time board model to discretize the vehicle's space-time trajectory; based on the space-time board model, a greedy conflict matrix decision algorithm is proposed. In order to improve the service efficiency of intersection and reduce the waste of time and space, a K translation optimization algorithm is proposed in this paper. To assist vehicles that could not be successfully reserved, try to avoid secondary requests and cross intersections as much as possible through acceleration or deceleration. This paper also extends the autonomous Intersection management of autonomous intersections, using two-way coupling simulator interconnection technology, using NS3 to carry out more realistic network transmission simulation to replace the simple network transmission model in AIM. In order to realize the simulation of autonomous intersection with feedback loop, this work is used to support the study of the demand for data transmission by different traffic control strategies and the effect of network transmission performance on traffic control efficiency. Finally, the control strategy proposed in this paper is simulated on the built simulation platform. The experimental results show that Batch-Light is superior to FCFS first Come First Service and traditional traffic signal control in both balanced and unbalanced traffic conditions. Moreover, the amount of data transmission in Batch-Light is small, with the increase of noise. Although the efficiency of service is decreased, the performance is stable, while the amount of FCFS data transmission is very large, and the influence of network performance on control efficiency is uncertain. Sometimes, with the increase of packet loss rate, the control efficiency will increase.
【學(xué)位授予單位】：大連理工大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：U491.54

【共引文獻(xiàn)】

相關(guān)期刊論文 前1條

1 王靜蓮;董相志;柳岸;;支持協(xié)同的工作流建模新方法[J];山東科學(xué);2007年04期

相關(guān)會(huì)議論文 前2條

1 黎明;劉偉兵;王先甲;;基于神經(jīng)網(wǎng)絡(luò)與群智能的多代理人決策與行為演化[A];第二十七屆中國控制會(huì)議論文集[C];2008年

2 張淑軍;孟慶春;;MAS中的個(gè)體學(xué)習(xí)與群體學(xué)習(xí)及在足球機(jī)器人仿真系統(tǒng)中的應(yīng)用[A];馬斯特杯2003年中國機(jī)器人大賽及研討會(huì)論文集[C];2003年

相關(guān)博士學(xué)位論文 前10條

1 趙麗寧;基于多AGENT的智能航運(yùn)信息系統(tǒng)關(guān)鍵技術(shù)研究[D];大連海事大學(xué);2010年

2 李慧;智能居住環(huán)境學(xué)習(xí)和控制策略研究[D];山東大學(xué);2011年

3 李s，

本文編號：1567894