1、Collaborative Research: Cyber-Enabled Demand-Interactive Transit (CREDIT) for Sustainable Transportation合作研究：基于互联网的已知需求的可持续定制公交1 IntroductionThe auto-dependent surface transportation system in the U.S. has not only created chronic traffic congestion that costs over $ 100 billion per year 109, but al
2、so severely limited mobility options, especially for those who have no access to private autos. Such a system also exacerbates the environment impacts of travel. In the US, transportation consumes 27.8% of all energy, of which 92.8% come from petroleum use; it also contributes 32.7% of greenhouse ga
3、s (GHG) emissions47.在美国路面上对于汽车的过度依赖的交通系统不仅造成了每年耗费1000亿美元的交通拥挤，而且严重的限制了群众出行的选择，尤其是对于那些还没有私家车的群众。这样的系统也恶化了交通的环境影响。在美国，交通耗费了总能源的27.8%，其中92.8% 来源于石油的使用，同时32.7%的温室气体的排放也由此产生。Promoting transit, or public transportation, is widely regarded an important measure in the transition to be more efficient, equita
4、ble and sustainable transportation 121.振兴交通，或者公共交通，被广泛的视为是实现有效、平等和可持续交通的重要措施。Being a mode of collective transport, transit tends to thrive in densely populated areas. In the US, unfortunately, suburb sprawl has made the condition progressively unfavorable to transit development since 1950s. In fact,
5、 the population density in the major urban areas has dropped more than 50% between 1950 and 1990 81, 69. Not surprisingly, transit has been largely marginalized across the country.作为一个公共交通方式，他会在人口稠密的地区茁壮的成长。不幸的是，自从1950年以来，美国散布的郊区使得其逐渐的不适宜于公共交通的发展。事实上，在1950年到1990年之间，主要城市的人口密度已经下降超过了50%。这样一来，公共交通被边缘化也
6、就不足为奇了。It is easy to see the dilemma faced by a traditional fixed-route transit service: on one hand, the potential ridership may not warrant the cost of providing high-quality coverage; on the other hand, the financially viable service is so inadequate that it attracts few non- captive riders.对于传统固
7、定路线公共交通所面临的窘境是显而易见的：一方面，潜在的乘客并不能保证提供高质量服务的费用；另一方面，在经济上可行的服务，并不足以吸引自由的乘客。To reverse the trend of declining transit use in the US requires a concert effort of planning, police-making and engineering. Indispensable in this effort is to transform the way by which transit systems are designed and operate
8、d. In order to succeed in lower density areas where traditional transit fails, the next-generation transit systems must better interact with passengers, more closely cater to their demands, and operate efficiently to minimize costs.为了扭转公共交通在美国衰退的趋势，需要规划，政策制定和工程的协力合作。设计和执行公共交通系统方面的改革是必不可少的。为了在先前公共交通失
9、败过的人口密度稀疏地区，继续运行公共交通，下一代的公共交通系统必须与乘客进行更好的交互，更加吻合于他们的需求，并且使用最小的成本达到最有效的运行。Technology is ready to enable such new systems. Many transit providers in the US now have the capability to track their entire fleets, make short-term projections, and distribute passenger information (on-line and on-road), all
10、 in real time.科技已经可以实现这样的新系统，现在美国许多公共交通供应商已经有能力监控他们的整个舰队，制定短期的计划，发布实时的乘客信息（在线在路上），所有的都在真实的时间里。With ubiquitously available Internet-enabled mobile devices 113, it has never been easier for passengers to request transit information and service, and to share their location and experience. Evolving towa
11、rds a smarter, demand-interactive transit system, however, calls for a holistic design approach, as well as the supporting analytical tools, to strike a delicate balance between system efficiency and user experience.随着无处不在的可以利用的能够上网的移动手机装置的发展，但是对于乘客要求获得公共交通信息和服务来，分享他们的位置与经验来说，这依旧相当困难。然而，交通系统朝着智能化与需求
12、交互式的方向发展，它需要一个整体设计方法,以及支持分析工具,从而实现系统效率和用户体验之间取得微妙的平衡。1.1 CREDIT overview1.1 交互式公交概述To the above end, we propose to investigate a new kind of transit system called CybeR-Enabled Demand-Interactive Transit, or CREDIT. CREDIT, as illustrated in Figure 1, can be viewed as a hybrid system that integrates
13、 fixed-route service (FRS) and demand-interactive service (DIS).根据前文可知，我们希望研究一种新型的被称之为基于互联网的需求已知的定制公交系统，下面简称“CREDIT”，如图一所示，“CREDIT”系统可以被视为一种固定路线服务（FRS）与需求定制服务（DIS）相结合的混合体。Specifically, DIS functions mainly as a connector between FRS and checkpoints close to passengersorigins and destinations. DIS di
14、rectly responds to demand in that passengers can only use this service through reservation. Unlike conventional demand-responsive transit (DRT), such as taxi and dial-a-ride, DIS neither offers a door-to-door service, nor guarantees all demands be fulfilled exactly as requested. The questions that w
15、e ask here are: Whether such a hybrid design could improve the performance of a transit system at no additional cost? If so, how much and under what conditions? How to configure a hybrid system to maximize the improvements? 特别的，DIS作为固定路线服务和与乘客起讫点相关的核查点相连接的功能。乘客仅需通过预留来使用这项服务，而DIS可以直接响应与乘客的预留。与传统的出租车等
16、这类需求响应的公共交通方式不同的是：DIS既不提供门到门的服务，也不能保证所有要求的需求都能得到充分的满足。那么我们会在这里提出这样的问题：在没有附加支出的情况下，这样的混合式设计能否提高公交系统的运行。如果是这样的话，花费多少并且在什么样的情况下才能够实现。如何形成这样的混合式系统从而使得改善最大化。The preliminary results presented in this proposal will offer a tentative yet positive answer to the first question the above. The key insight is th
17、at DIS would provide a low-cost enhancement of spatial and temporal accessibility such that FRS can be operated more efficiently, e.g. by more widely spaced lines of higher frequency.对于上面提出的第一个问题，根据这个提案所显示的初步结果将提供一个实验性的但是积极的回答。主要的观点是DIS将提供一个对于时空效率的低成本加强，这样的话FRS可以更加有效率的运行。比如说：增加高频率的空间线路。The main char
18、acteristics of CREDIT follow.CREDIT的主要特点如下： Passengers may request to board and alight at specified checkpoints through a mobile transit application (e.g. a smartphone app), see Figure 1. The application will present one or more trip options generated by the trip planner. These options may have diff
19、erent pickup checkpoints, starting times, paths, and prices. Once the passenger selects an option, thefare is paid and the system commits to the selected service.乘客或许通过一个通信交通应用（比如说一个智能手机的应用）要求在具体的核查点上下车。该应用通过出行规划产生一个或者多个出行选择。这些选择拥有不同的临时组合而成的核查点，开始时间，路径，和价格。当乘客挑选了一个 选择，费用就被支付了，并且提供所选的服务。 Each on-dema
20、nd vehicle is assigned to feed one fixed-route transit line or a few lines located close to each other. It travels through the regular stops of the lines and the checkpoints between them to pick up or drop off passengers. The exact itinerary is generated by on-demand router according to the confirme
21、d requests. To be efficient, each vehicles service area is restricted to a narrow neighborhood of the assigned lines.一条固定的公交线路或者几条邻近的公交线路将被分配给每一辆被定制的车辆。它会在路线上的几个常规车站和核查点上上下乘客。确切的行驶路线由路线使用者确定的要求进行定制产生。为了确保效率，每一辆机动车的服务区域均被限定在被分配线路的邻近区域。 Real-time information, such as location and projected arrival tim
22、es of any transit vehicles, is collected and distributed via displays at stops, web portals and mobile applications. The trip planner will continuously monitor such information and update the trip guidance throughout the journey.通过车站的显示牌，网络的出入口和通信设备的应用搜集和发布实时的信息，比如说任何一辆公共交通车辆的位置信息，具体的到达时间等。出行规划者在整个行
23、程中不断地监视这些信息，并不断更新出行指南。1.2 Research objectives1.2 研究对象The proposed research aims to create theories and tools to characterize, analyze, operationalize, and evaluate CREDIT. The research activities are organized under three thrusts, each addressing a specific objective, as described below.提议的研究目标是产生如何
24、描述这些特征，分析，运营和评估CREDIT的理论与工具。如下所述，研究的范围限定在三个方面，其中每一个都致力于确切的对象。The first thrust addresses the characterization and strategic design of CREDIT. The overarching goal here is to test and verify the hypothesis that a hybrid system like CREDIT can improve user experience at small or no additional costs for
25、 a wide range of demand scenarios. 第一点致力于CREDIT的特征描述与战略上的设计。本文的更高层的目标是测试并证实如下的假设：对于较大范围的需求的情况下，较少的或零附加成本的像CREDIT这样的混合系统可以提高用户的体验To understand the working mechanisms of CREDIT, conceptual models of the system will be built using the continuous approximation approach. The simplicity of this approach w
26、ill allow close examination of a variety of fundamental issues, ranging from hierarchical route structure to FRS/DIS tradeoff and the impact of demand adaptation/coordination. Insights drawn from the findings will guide tactics design and operations in the second thrust. 为了理解CREDIT的工作机理，该系统概念模型是基于近似
27、连续方式进行建立的。这个方法的简化将允许对于从出行者路线结构到FRS/DIS的交易和需求适应性/同等性的影响等这些不同的基础数据的封闭测试，从研究结果中提出的观点在第二点中将要指导战略的设计和运行。The second thrust formulates, analyzes and solve mathematical problems that support operations. The focus will be given to three core decision modules in the system, namely on-demand router, tactics de
28、signer and trip planner (see Figure 1). Specifically, on-demand router plans vehicle itineraries based on confirmed requests; tactics designer coordinate schedules and adapt services to demands; and trip planner generates optimal trip guidance according to available services and historical and real
29、time information. 第二点致力于公式化表达，分析并解决支持运行的数学问题。该系统的焦点将放在三个核心的决策模块上：即路线定制者，策略设计者和出行规划者（参看图一）。特别的，路线制定者根据确定的要求安排机动车的旅程，策略设计者协调时间表并使服务满足于需求。出行规划者根据可以得到的服务和历史与实时信息产生最优的出行建议。As these problems vary in complexity and spatial/temporal scales, they will be addressed by different modeling techniques. Efficient
30、algorithms will be studied and implemented to solve the above problems, subject to real design and operational constraints.由于这些问题复杂性和时空测量的不同，他们将通过不同的模型技术进行解决。有效的算法将要被研究和实施去解决上面的问题，旨在真实的设计与运行的限制。The last thrust evaluates the proposed system and tools. The evaluation framework will havetwo pillars: a
31、discrete-event simulation platform that implements virtual CREDIT systems, and a smartphone-based real-time transit trip planning application known as Transit Genie. Traffic supply and demand data will be obtained from open sources and local transit agencies and synthesized to create case studies. D
32、ifferent system designs and operations policies will be evaluated using these case studies on the simulation platform. 最后一点致力于评估提出的系统和工具。评估的框架有两个支柱：一个独立事件的仿真平台运行虚拟的CREDIT系统，和一个众所周知的基于智能手机和实时的公共交通出行规划的应用：Transit Genie。交通中提供和要求的数据来源于开放的资源和当地的交通局，并经过综合产生案例研究。在仿真平台中不同的系统设计和运行政策将要通过这些案例进行评估。Transit Genie
33、 will be enhanced with the ability to collect de-identified location traces from its users, which offers access to a potentially large source of transit demand data that does not currently exist. These GPS traces will be first processed through semantic information recovery into meaningful itinerari
34、es, and then compared to optimal itineraries produced by virtual CREDIT for the same trip.Transit Genie将要加强它收集确定用户的位置轨迹的能力，这将挖掘出现在并不存在但是拥有巨大潜在的公共交通需求数据。对于同一个旅程，这些GPS轨迹首先经过语义信息转换为有意义的旅程，然后与虚拟的CREDIT产生的最理想的旅程进行比较。1.3 Intellectual merits1.3 研究价值This work systematically explores the integrated design an
35、d operations of a hybrid transit system enabled by communication, computing and location technologies. By proposing, analyzing and evaluating CREDIT, the project will test the hypothesis that a properly designed hybrid system can both improve user experience and be cost-efficient.本工作系统的探索了能够交流，计算并拥有
36、定位技术的混合式公共交通的整体设计和运行。通过提议，分析和评估CREDIT，本项目将要验证如下的假说：对于混合式系统的设计既可以提高用户的体验，又可以使得花费最有效。It promises new theories to explain the working mechanisms of such hybrid systems and to guide their strategic designs. The research will also create a suite of mathematical models and algorithms for tackling challeng
37、ing operations problems arising from CREDIT. The knowledge and insights gained from this project will not only benefit the design and implementation of next-generation transit systems, but also contribute to the understanding of other cyber-physical systems.它允许新的理论去解释这个混合系统的工作机制，然后去指导它的策略设计。本研究也会产生一
38、套数学模型和算法，用以解决由CREDIT引出的具有挑战性的运行问题。从这个项目中获得的知识和观点不仅会使得下一代的公共交通的设计和运行收益，而且对于其他的物理网络系统的理解做出贡献。Finally, a comprehensive evaluation framework based on simulation and driven by real-world data will be developed to evaluate various CREDIT systems, as well as the analysis tools.最后，一个基于真实世界数据驱动的仿真综合的评价框架将要发展
39、起来用以评估不同的CREDIT系统，作为一个分析工具。2 Related studies2 相关的研究The focus of this section is to review previous proposals to combine demand-responsive transit (DRT) with fixed-route services and to compare them to CREDIT. It starts with a brief introduction to DRT itself, which has been the subject of intensive
40、research efforts since 1970s. Also included here are the related works on transit network design.本节的焦点是审查先前关于结合需求响应交通（DRT）和固定路线交通（FRT）的提议并与基于互联网的定制式公交（CREDIT）进行比较。它开始于对于DRT的简单介绍，DRT从1970年开始就已经成为研究的热点。这里也包括对于公交网络设计的相关工作。2.1 Demand responsive transit2.1 需求响应交通Of all forms of DRT, the most relevant to
41、the proposed work is dial-a-ride transit (DART). In a DART system, a passenger requests a service and is provided a pickup time window. Unlike taxi service, DART allows ride-sharing and even transferring 124, 117. DART may be implemented as an immediate response transportation mode or a subscribed s
42、ervice through advance scheduling 119, 45. 在所有的DRT方式中，与本文工作最相关的就是电话预定交通。在电话预定的交通系统中，乘客要求一个服务，然后被提供一个选择时间窗口。与出租车服务不同的是，电话预定交通允许合乘，甚至是换乘。电话预定交通或许被作为一个迅速响应的交通方式进行运行，或者作为一个通过先进的时间表实现的预约服务。Dynamic ride-sharing, which has received much attention recently 58, 43, 1, 63, aims to match travelers with similar
43、 itineraries and schedules on short-notice. While it represents a very different form of public transportation, dynamic ride-sharing often faces similar optimization challenges as DART 2.高效的合乘目前受到了更多的关注，它的目的是对有近似旅程的乘客进行匹配并在时间表上作简短的提示。当它作为一种非常不同的公共交通形式时，高效的合乘经常面临作为电话预定交通的最优化的挑战。The basic mathematical
44、 problem underlying DART is a many-to-many pickup and deliveryproblem with time windows, which may be further classified as single- and multi-vehicle problems depending on the number of vehicles employed, or static and dynamic problems depending on whether or not the passenger requests are processed
45、 in real time. The DART problem is a special class of vehicle routing problems (VRP) that are known to be NP-hard . 隐藏在电话预定交通下的基本的数学问题是多对多的匹配和时间窗口的运送问题，根据雇佣车辆的数量可以被更深层次的分为一个或者多个机动车的问题；根据乘客要求的实时进程可以分为静态与动态的问题。电话预定交通的问题是一类特殊的机动车路线难题，即众所周知的NP 难题。Many attempts have been made to develop efficient DART so
46、lution methods in the past three decades. Exact algorithms, such as those based on dynamic programming 99, 100, 44 and branch-and cut 26, 107, 106, are only effective in solving relatively small instances. Not surprisingly, much more efforts have been devoted to searching efficient heuristics, rangi
47、ng from simple insertion techniques 67, 3, 46, to more complicated decomposition schemes and meta heuristics.在过去的30年里，许多尝试都已经做了去发展有效的DART解决方案。确切的算法，比如说基于动态规划和分支定界法和割平面法等，仅对于解决相关的一些小型案例有效。并不奇怪，从一些简单的插入科技中排列，更多的努力被贡献于研究有效的启发式算法。Another line of analysis approaches the complex DART problem through conti
48、nuous approximation (CA) 33. The focus of such analysis is to construct optimal routing strategies based on asymptotic behavior of traveling salesman tours 51, 9, 116, 39. The problem is typically solved by first partitioning the service area into zones (or clusters) and then constructing within-zon
49、e tours 117, 38, 85, 86, 17, 71.另外一条研究复杂的DART难题的方法是通过近似连续的方法。这种分析的焦点是基于旅行商的渐近线行为构成最优的路线策略。这个问题被第一分割服务地区为区域而解决了，然后形成了区域的旅行路线。2.2 Hybrid transit systems2.2 混合的交通系统A major shortcoming of DRT services such as DART is the high per-capital operating cost resulted from low occupancies 16. This observation has inspired proposals of hybrid systems that strive to marry the flexibility of DRT with the efficiency of traditional fixed-route services. In fact,many transit agencies in the US are practising some forms of hybrid operations, ranging from route/point deviatio
