Skip to main content

MoveWise-Research project

Periodic Reporting for period 1 - MW-R (MoveWise-Research project)

Reporting period: 2017-09-29 to 2018-09-28

The key objective of the MW-R project is to exploit the capabilities of user interaction to deliver more accurate information to its users through a better understanding of the mechanics of the transportation system. Transport Network operators (e.g. train operators, bus operators) require a significant amount of data to perform their planning operations. Those data are usually disconnected from the realistic incidents. For example, imagine some works being performed at a parking lot that serves a train station. Through the data, the operator can see the reduced amount of inflow to the station and might even know that the parking lot is closed. However, this event is not quantified in the database which results in two problems; first, the traveller does not know the impact of that event on his/her travel time and second, the operator cannot relate the incident with statistical analyses on its data without making assumptions (assumptions on the impact of the event).
MW-R was created to analyse and provide quantification methodologies of the impact of various events on the network utilizing user feedback (crow sensing) and data assimilation techniques. MW-R aims to provide improved and more accurate information to the travelers and also enrich the datasets of operators thus, reducing the cost of planning operations and improving the power of those operations. MW-R assists in delivering a better planned and more inclusive Public Transport operational scheme.
MW-R set out to create the foundation for a complicated product. The innovation associate was tasked with creating the architecture of the product’s operation, conceptualise the main functionalities and create the first part of the product’s algorithm (namely, the data assimilation component). All the aforementioned goals were achieved through various steps. Specifically, the associate initially created the architecture of the product by performing literature review. After creating the architecture, the associate created an online questionnaire survey which aimed at researching interaction with a “prototype” of a MoveWise application. The survey provided a problem to the surveyees/responders and requested them to acquire information by selecting one of three options. The survey provided interesting information concerning how respondents will tackle similar feedback requests from the actual MW-R application and yielded the idea of including the level of certainty as one of the fields in a feedback within the application. The associate then proceeded with creating a demonstration of the envisioned application. More specifically, the demo aims to show how the data assimilation component will operate using real data. The demo aims to show the conceptualized functionalities can be used in real world situations; the model can predict the network flows in short time span, the model can detect “triggers” that generate feedback request and that the feedback results in a different simulation result that, ideally and if the feedback is accurate, is closer to the actual measurements.
The demo and the work performed has been accepted and will be presented in Deep Dive Lisbon (16-19/10/2018), a meeting that aims to connect startups with investors. Specifically, the Deep Dive week is aimed at analytics and data startups “ready to scaleup”. Furthermore, AETHON is in contact with a transport operator in Portugal with the goal of obtaining new data and investment for MoveWise next cycles and for providing an operational version of the product for the Portuguese Public Transport. AETHON is actively looking for synergies that can accelerate the product while also attempting to use the algorithm as is for ad-hoc/procurement type work (i.e. off-line assimilation of various datasets for PT operators).
The project has already been presented at April 2018 in Transportation Research Arena, Vienna and is scheduled to be presented in MATTS 2018 conference at October 2018. Furthermore, the work has been presented in an open publication at arXiv (Cornell University Library - and will be published in a scientific journal following a successful peer review within 2018 – beginning of 2019.
The data assimilation algorithm developed during the MW-R project implements the Kalman Filter model in a novel way as far as the transportation engineering community is concerned due to the implementation of the control term. Furthermore, the model has potential for rapid implementation thanks to employment of a standard model for which software libraries are readily available.
The impact of the project is related first with the operators. The project aims to provide a significant increase in the power of the data that the operators use for performing planning activities thus increasing the efficiency of the planning activities and, consequently, increasing the profit of operations. By performing better planning operations, MW-R contributes in creating a better network and service for the passenger all the while including crowd sourced data. Thus, MW-R is an enriching platform that improves the power and usage of existing datasets, by performing fusion/assimilation and filtering. However, MW-R is not only targeted to operators. It is also targeted to the passenger that is on the road. Specifically, MW-R aims to use the short-term space model that is used for data assimilation, in order to provide improved recommendations to passengers. In simple terms, the feedback is inserted in the simulation engine (state space model) and achieves improved estimations in the short term (15 min interval). This estimation is aimed to be transmitted to passengers in order for them to plan their trip better. Imagine that MW-R can take the form that “rain-alarm” take for weather prediction apps. If a problem is detected in the short-term, MW-R generates a feedback request, creates a new prediction and disseminates it to passengers.
MW-R achieved some indirect impacts as a project which relate with the hire of an experienced Innovation Associate that excelled in lifting the idea off and in providing useful information to the company by sharing knowledge retrieved in different working environments and in different fields (academia and industry). This knowledge and in combination with the INNOSUP work plan assisted the company in creating a new education scheme for the employees where each employee can take 4 days per 6 months to participate in a course or education seminar for which the company provides vouchers.
Finally, the company and the innovation associate have created plans for the future of MW-R. The created model/algo can be implemented through different business models, either as a standalone application or as an API connected with existing application, thus, working as an ad/subscription-based app or as a pay-per-call API app. Both models will be examined during the next research cycle which will be coupled by business analytics.