Dynamic vanpool services: Passenger preferences, operations modeling and simulation-based quantification of impacts

在这个项目中，我们建议从乘客，运营商以及政策制定者的角度来研究与动态vanpooling相关的基本问题。具体而言，我们希望（i）量化乘客对动态拼车的各种服务属性的偏好，如接送延迟，路线的迂回和票价;（ii）建立一个仿真平台，可以让我们定量评估动态拼车的性能及其社会影响;（iii）开发一个高性能的调度算法，这是动态拼车的核心技术。它应能够处理需求和出行时间的随机性，并能够协调上海等特大城市的车队;以及（iv）评估动态拼车的社会和环境影响，并确定如何将其与现有出行方式相结合，以满足居民的出行需求。该项目的成果将使我们能够全面了解乘客的偏好、动态拼车的运营挑战及其对建设更环保、更可持续的城市的影响。该项目的总体方法论方法结合了四个相互关联的工作包（WP）。第一个工作包包括乘客偏好模型的估计，这是必要的下游任务的研究项目。特别是，我们从熊猫巴士收集了陈述偏好（基于专门电子问卷的智能手机调查数据）和揭示偏好数据，并将其与智能手机和车辆上的传感器数据相结合。然后，我们使用这个工作包的输出作为工作包2和3的输入。在WP 2中，我们扩展了交通建模和仿真系统的最新技术，以便它们可以捕获动态的vanpooling服务。在WP 3中，我们将WP 1的输出和开发算法的服务车辆的最佳调度，在一种方式，最大限度地提高服务质量为给定的车队规模。在工作包2和工作包3中，我们都明确地处理了需求和供给的不确定性。在WP 4中，我们使用工作包2和3的输出，并开发了一个基于模拟的敏感性分析框架，用于对vanpooling服务的影响进行网络范围的分析。基于模拟的敏感性分析框架优于对少数不同网络的影响分析，因为其结果可以概括为每个影响与每个模型输入之间的关系。例如，我们的模型的输出将是一个特定的公式，将需求水平、网络类型和交通组合与每种类型的影响水平联系起来。因此，结果将可用于任何类型的vanpool服务，而不限于分析中的特定系统。

In this project, we propose to investigate fundamental problems associated with dynamic vanpooling from the perspectives of passengers, operators, as well as policy makers. Specifically, we want to (i) quantify passengers preference toward various service attributes of dynamic vanpooling such as pick-up delay, circuity of the routes, and fare; (ii) build a simulation platform that can allow us to quantitatively evaluate the performance of dynamic vanpooling and its social impacts; (iii) develop a high-performance scheduling algorithm, which is a core technology for dynamic vanpooling. It shall be capable of handling stochasticity in both demand and travel time, and can coordinate a fleet of vehicles in megacities like Shanghai; and (iv) assess the social and environmental impacts of dynamic vanpooling and identify ways to integrate it with existing travel modes to serve the mobility needs of the residents. The outcome of the proposed project shall allow us to develop a thorough understanding of the preferences of the passengers, the operational challenges of dynamic vanpooling, and its impact to building greener and more sustainable cities.The overall methodological approach for the project combines four interrelated work-packages (WPs). The first work-package covers the estimation of passenger preference models, that are necessary for the downstream tasks of this research project. In particular, we collect stated preference (smartphone-based survey data based on a specialized electronic questionnaire) and revealed preference data from Panda Bus, and couple them with data from sensors on the smartphones and the vehicles. We then use the outputs of this work-package as inputs to work-packages 2 and 3. In WP2 we extend the state-of-the-art in traffic modeling and simulation systems, so that they can capture dynamic vanpooling services. In WP3, we incorporate the outputs of WP1 and develop algorithms for optimal scheduling of the service vehicles, in a way that maximizes the service quality for a given fleet size. In both work-packages 2 and 3, we explicitly deal with demand and supply uncertainty. In WP4, we use the output of work-packages 2 and 3 and develop a simulation-based sensitivity analysis framework for the network-wide analysis of the impacts of the vanpooling service. The simulation-based sensitivity analysis framework is preferred over the analysis of impacts on a small number of different networks, as its results can be generalized into relationships between each impact and each of the model inputs. For example, the output of our model will be a specific formula linking the demand level, the type of network, and the traffic mix with each type of impact level. Thus, the results will be useable by any type of vanpool service, and not restricted to the specific systems under analysis.