After three years of experimentation, research and development, the startup Padam Mobility offers transport operators its solutions to transform bus services. Thanks to artificial intelligence, the lines, schedules and routes are now created and optimized in real time while user experience is completely renewed.
Mobility and the difficult equation in low- and moderately dense areas
Padam Mobility’s project was born from simple observations: on one hand, a rigid, not reliable and unsuitable to the people’s needs create a strong frustration of users; on the other hand, a way of bus operating that has not evolved for decades despite new digital tools available (smartphone, cloud, real-time optimization). In all circumstances, once fixed lines are drawn with fixed schedules, supply is no longer adapted to demand, which leads to many under-optimizations.
The territories must solve an extremely difficult economic and environmental equation for mobility. Except in a few major metropolises, the modal share of the car does not decrease despite a widespread political will. A significant effort is also still to be made for inclusive mobility to ensure the basic mobility needs of those who do not own a car. But to tackle these environmental and social imperatives, local authorities must contend with constant or decreasing budgets.
This tends to cast a harsh light on the under-optimization of public transport services. It is no longer possible to see, day after day, vehicles with fifty seats driving almost empty, while a demand for mobility remains unserved. This explains in particular the craze in recent years for carpooling, bicycles – electric or not – and other micro-mobility solutions. Despite these new offers, the main challenge is to improve public transport, especially buses, which are the bedrock of mobility in all territories. This is where we will find a performance optimization improving access to mobility and reduce the cost per trip.
That’s why we created Padam Mobility with the ambition of transforming the business of bus operator through artificial intelligence. On the principle of DRT, we first experimented shared shuttles in Paris region, by night, and then on daily mobility from home to work. In 2017, we transformed this Software As A Service products for transport operators.
Transforming the trade of transport operator with artificial intelligence
A first wave of on-demand transportation services was introduced in the 1990s. These reservation services (by telephone) have made it possible to serve some territories in a very light way, but the ambition of these services has so far been limited and the tools to operate them insufficient to be transposed to transport Mass.
We have entered a whole new era since we have the ability to know and manage services entirely in real time. The Padam Live operating platform© is the embodiment of this idea: customers book and follow their journey on their mobile application, drivers follow a roadmap constantly updated on their tablet, operators know to at any time what is the state of the fleet and the system as a whole. For truly real-time management, the challenge is of course to make decisions in a fully automated way on the allocation of vehicles, the calculation of their route and the information traveler, which is unique to intelligence Artificial!
Technically, this level of automation presents several challenges.
From a scientific point of view first: when we started the project, we realized that very little work in the literature concerned the optimization under constraints and real time of a fleet of vehicles, the “online” methods. Until now, so-called VRP (Vehicle Routing Problems) problems were always handled by “offline” methods, before the smartphone and cloud gave relevance to a new generation of systems optimization. So we took these problems at the base with experts in Operational Research and Optimization under Constraints, without recovering any existing bricks.
Second, automation requires a strong operational experience. By nature, transport faces many external constraints. To name but two: user behaviour (customers or drivers) can cause disruption at any time (delays, errors, change of mind, etc.) and congestion adds a strong random component. Only a good operational experience can lead to a high level of automation and reliability, which justifies the experiments that Padam conducted in its early days, as other market players have done.
Mobility and artificial intelligence: what are the prospects?
The stakes are up to the operational technical difficulties. As these begin to be overcome, we see radical performance gains on the projects we are working on. To serve a fixed mobility demand, on-demand transport can reduce costs by an average of 30 per cent compared to conventional fixed-line bus services.
Excluding highly charged and therefore efficient fixed bus routes, we estimate that in the long run forty per cent of bus journeys will be made on an “on-demand” mode. All public transport networks are affected by this transformation, from rural areas to the outskirts of major metropolises. Beyond the pure performance gains, the real-time management of the services allows a better response to hazards (failures, road problems) and therefore a much better reliability. The customer experience is transformed, it is accompanied from the search of the route, until its arrival at its destination with a quality of information-traveller unprecedented.
Finally, it is difficult not to put the development of on-demand transport in perspective of that of autonomous vehicles. These will be expensive to purchase but with a lower cost per kilometre than a driver-driving vehicle, which means that transportation services will make the most of it. The interface with on-demand transportation services will be natural. We can even risk a prognosis: just like on-demand transportation, it is first in medium-sized cities that autonomous vehicles will be truly adopted. The constraints , especially roads, are less complex and the economic stakes of mobility are the strongest.
Article originally published on Telecom ParisTech
