Mobility is a fundamental pillar of the economic activity, the territorial activity, and people’s daily lives. Today, the mobility system is facing many challenges:
Environmental transition and climate change: decreasing mobility carbon footprint and preserving the air quality,
Resources and budget allocation: the increasing scarcity of public budget is leading to optimizing investments in the transportation infrastructures,
Technological transformation and new uses: new means of transportation requiring new infrastructures (micro-mobility, shared mobility…).
These three transformations cause large scale issues for public authorities in charge of mobility, for transportation infrastructures operators, and for mobility operators. Decision-making must be guided beforehand to optimize the allocation of resources. A “data-driven” vision has become necessary to bring optimal guidance.
Mobility data as a door opener
It is no news: data is the new black gold. It is part of our everyday lives, produced regularly in mass by thousands of apps and systems to improve our lives. It is fueling all domains of activity and particularly the mobility sector.
The first question that comes to mind is: how is the mobility data collected? Automatic counts (loops, videos, radars, …), localized surveys and GPS and smartphones terminals are common data sources to estimate traffic flows and trips volumes. Data collected from connected vehicles and smartphones (known as FCD or FMD) enables to estimate travel times and flow speed. Mobility carbon footprint accounting is made possible with a fine knowledge of the car fleet and of the pollution caused by each type of vehicles. In other terms, data is a major technological asset to depict real life phenomena. Data is a door opener to bring informed guidance on transformation strategies and to tackle mobility challenges.
Optimizing mobility system and mobility flows with data is a process that can be based on two different approaches: statistics and modeling. The statistics approach is about mobility data processing with technologies and methods coming from data science and artificial intelligence (machine learning, for example). It focuses on historical datasets to dig out key indicators and generates predictions on eventual future situations. The modeling approach is about building a model of the mobility system recreating the entire interaction between the offer (network) and the demand (traffic flows and people trips). This model recreates virtually the whole flows in the mobility network and is calibrated to a referent situation, built thanks to a partial knowledge of the said situation.
Artificial intelligence and simulation digital twin, two different processes
Regarding the statistical approach, once the data is collected and provided, it is about processing and analyzing it to dig out relevant information. This information is then classified to show occurrences of similar phenomena. This classification relies on artificial intelligence and machine learning technologies to highlight multiple similar and recurring phenomena, from which indicators are picked up to support the decision-making. The statistical approach not only help to understand the causes of a phenomenon, but also its consequences. It is about projecting future situations from an observed one at present time and from the understanding of the past occurrences.
However, it is important to nuance the benefits of the statistical approach as its strength depends on the recurrence of the observed phenomenon: the more the phenomenon happens, the more relevant the statistics are and the more substantial they will be to the decision-making.
With the simulation digital twin approach, a digital model of the mobility system is built with available data sets and reproduces static or dynamic flows in the network (road traffic for example) to simulate a real phenomenon in a virtual environment and evaluate it in different scenarios. This digital representation of a physical reality enables to simulate scenarios which would require overly high financial investments (reorganizing crossroads for example) or logistics too complex (a special traffic management measure) to be tested in the real world before their deployment. We are then able to model and simulate traffic management operations to evaluate their operational efficiency and optimize them if necessary. The digital twin can predict situations even if they were never observed in the past, and this is its strength compared to the statistical approach. For example, one would be able to simulate the impact of an accident on a given traffic lane, even though this phenomenon never occurred in the past. With mapping visualization and dashboards, the simulation digital twin provides easy-to-use tools and key insights in a cloud-operated platform.
Building synergy with the two approaches to better fit the needs
Though these two approaches bring guidance on their own, they won’t be as relevant according to the question to answer: if the need is to understand causes and consequences of recurrent and predictable phenomena, the artificial intelligence approach will be bringing accurate and appropriate answers. On the other hand, if one wants to predict consequences from a phenomenon that never happened before and of which we have no data on similar situations, the digital twin will be more efficient. Nevertheless, the key point to maximize the benefits of these two approaches is the synergy they offer when used in tandem. It is still a very innovative practice on the market today.
And yet, easy-to-use software solutions are available on the market. They knock down the technological barriers by combining a data aggregator module, artificial intelligence-based tools, modeling, simulating, and predicting engines with advanced computing and collaborative capabilities of cloud-based architectures.
Wish to discover how Neovya is guiding the decision-making by using these two approaches?
Talk with an expert ➜ Book a meeting!