Urban mobility is undergoing unprecedented change. New technologies and business models are challenging the traditional and current organisation of transport systems. New mobility services (shared mobility, micromobility, MaaS or electric mobility) and behavioural changes are part of this revolution. Transport modelling is essential to understand, monitor and anticipate these changes. However, standard modelling approaches (so-called "four-stage" models) are ill-suited to this new context, making it difficult to implement comprehensive, relevant urban transport planning at a time when urban transport is facing a multitude of challenges: The use of multi-agent models for urban mobility planning aims to provide a better understanding of all transport modes, especially those that operate on demand, such as micro-mobility, mobility as a service (MaaS), car-sharing and autonomous vehicles.
State of the art, current situation
The four-phase models used in most French conurbations have proven their relevance for planning road and public transport networks (metro, tramway, bus). However, because of their aggregated approach (in space, time and behaviour) and their focus on peak periods, these models are no longer relevant for accurately describing the diversity of behaviour and interactions between users and modes of transport. For example, these models do not allow a self-service bicycle or car-sharing fleet to be satisfactorily modelled and sized. This is problematic for public policy makers at a time when behaviour is increasingly multi-modal and multi-rhythmic, varying from day to day and time of day.
Methods, technical ideas, innovation
Within the framework of the Egis x ENTPE chair to be signed in 2021, the LAET and Egis have developed a multi-agent decision support model for the Lyon conurbation in order to respond to relevant use cases for local stakeholders. The modelling platform chosen is based on the MATSIM open platform, a free tool widely used in the scientific community. The aim of this collaboration between a renowned research laboratory and a leading consultancy firm is to make these disaggregated modelling techniques, which have so far been confined to the world of research, available to public decision-makers and regional planning. In this way, the innovation lies not in the development of a model, but in its deployment and use in public policy making.
Theoretical or experimental results, interpretations
This is an innovative modelling tool that simulates a population, represented by agents, in their daily activities. For example, in the Lyon Metropolitan Area model, 1/4 of the city's population is simulated over a 24-hour period. Each person is represented by an agent who has activities to perform and uses different modes of transport to carry them out. In this way, the level of use of each mode of transport is determined by the activities carried out by the agents.
Egis, through its partnership with LAET, has a multi-agent model for the Lyon Metropolitan Area which has already been used to carry out initial modelling on three use cases:
- Theoretical sizing of the scooter fleet in the city of Lyon.
- An analysis of the potential for extending the Vélo'V shared bike system.
- An assessment of the impact of parking regulation policies on shared micromobility in Lyon.
Initial results: the case of Lyon
Deployment and way forward
In addition to the cases presented, other use cases have been identified and will be tested. In particular, the issue of freight transport will be addressed. Egis and the LAET are always on the lookout for new use cases and new territories interested in experimenting with multi-agent models. Do not hesitate to contact us.