Case Studies

Data Science on Road operations

Background

Today we live in a world, where objects and people are getting more and more connected. Road infrastructure operators are implementing different types of sensors, they inform us about weather conditions, level of pollution, noise level, energy consumption, traffic flow, etc. Besides these traditional sensors, the mobility related data is drastically enlarging, by means of emerging vehicle technologies (connected and autonomous vehicles are discussed all over the globe) and, above all, communication and mobile technologies (every user is now a mobile sensor thanks to smartphones and connected devices).

The interconnection of all these data makes it possible to have a detailed knowledge of the state of a vast vehicle-infrastructure-user system. Road network operations and traffic control will be impacted because it is now a complex system that we want to regulate to make it environmentally sustainable (by minimizing pollution, waste, energy usage, noise), safe (by monitoring and preventing accidents, maintenance, natural disasters) and valuable to users (by reducing travel times and improving the quality of experience).

On a technical level, we have new challenges to deal with, huge volumes of data generated by this multitude of sources that are often heterogeneous, asynchronous and whose reliability is not always demonstrated. Internet of Things and connected devices provide the basis for Big Data analytics. Big Data techniques provide some answers. Artificial Intelligence is needed to process these big amounts of data.

Jointly with a Southern Europe road construction and maintenance enterprise, we developed a methodology for driving its transformation program into a data-driven Company capable of exploiting all the new opportunities deriving from this new and continuously evolving scenario. More than 50 use cases were identified with a comprehensive approach across the operations department, leveraging on cutting-edge Big Data & AI technologies – such as Machine Learning, Deep Learning and IoT Analytics - to exploit the heterogenous and extremely valuable data asset of the Company.

Decisions and actions

The journey toward a data-driven company has been achieved through the execution of Bip Data Strategy program, aimed at achieving the following objectives:

  • Determine the current state of the Client’s business, through Bip’s Data Strategy Maturity Assessment;

  • Define a vision and the path to achieve it with guidelines on the four dimensions: strategy, organization, analytics and technology;

  • Design the short-medium term roadmap including a set of use cases to be implemented with an agile approach and short-term actions identified to start covering basic gaps on the four dimensions.

The main actions taken can be summarized as follows:

  • Organization: In 2018 a series Big Data & Artificial Intelligence workshops were conducted with the participation of the corporate top management at first, and then involving the mid-management and employee. The purpose of the workshops was to engage the client on new Big Data attitude in order to become a data-driven company, and to focus on IT Big data infrastructure.

  • Technology: The first prototypes of the use cases were realized on-premises using simple Virtual Machine and Python in late 2017. In July 2018 a Big Data Platform has been implemented on Cloud; in 2019 AIOps methodology has been put in place for Artificial Intelligence development and deploy.

  • Advanced analytics: In two years, several Advanced analytics use cases have been tested and deployed into production. These involved mainly:

    • statistical analysis, such as analysis of typical traffic conditions, based on the past, and real-time traffic conditions;

    • forecasting models, short-term and long-term prediction of traffic speeds, also based on the impact of external factors;

    • what-if analysis and simulation of the effect of external factors on typical traffic conditions;

    • cognitive models, such as text analytics for road infrastructure database reconstruction and cognitive search engine on road infrastructure archive;

    • custom algorithms to identify the most risky road asset items based on their structural condition in order to optimize the extraordinary maintenance interventions.

Results

The development and successful implementation of the Data strategy Roadmap over the two years has brought the following benefits to the client:

  • Creation of a unique tool for monitoring and supporting decisions through the processing of large amounts of data and AI enabled scenario simulation;

  • A set of Machine Learning models automatically trained over regular intervals of time, updated with the most recent traffic information, to refresh the estimates maintaining low error in in the predictions for over 78000 road segments all over the country;

  • Development of a management system for programming road asset maintenance based on structural engineering analytics;

  • Complete view on road asset condition state, maintenance needs, engineering inspection results, asset geo-localization on maps through a user-friendly integrated interface;

  • A Big Data Platform on Cloud is available to store all relevant mobility data, to support the development of future AI application and the change toward a proactive and data-driven attitude toward business decisions;

  • Promotion of the awareness on Big Data and Advanced Analytics at enterprise level.

Get in touch

Are you ready to make sense and make things?