Forward thinkers at DTU High Tech Summit 2019: Professor Francisco Pereira, DTU Management. “The high complexity of the transport system makes it ideal for applicating Machine Learning. Machines can identify the patterns of movements and also transform them into predictions of future movements much more accurately than even the most skilled human observer would be able to.”
Professor Francisco Pereira, DTU Management
Francisco Pereira heads the DTU research group Machine Learning for Smart Mobility.
“The transport sector is behind other sectors when it comes to sustainability. Finding more sustainable mobility solutions is thus a strong societal need. A flexible and adaptive system based on Machine Learning will make public transportation cheaper and more efficient, and thus more attractive compared to driving your own car. This is a huge advantage for the environment and for climate protection.”
Watch out for adverse effects
While a transport system based on Machine Learning promises improved sustainability, developers need to be alert to possible adverse effects:
“Suppose that the system due to an error predicts a lot of people in need of transportation at a given location and sends a large number of vehicles to that spot in vain. This would be sad. Firstly, we would have congestion at that particular location, and secondly, contrary to the overall idea, rather than improving sustainability we would damage it by circulating empty vehicles.”
A similar challenge is things changing in the physical world:
“The system may have planned everything optimally, but if a major accident or some other unforeseen event happens the situation can suddenly become totally different. Therefore, the system needs to be not only efficient but also resilient.”
Smooth traffic, easier lives
Not only will Smart Mobility increase the sustainability of transport. It will also make our lives a lot easier, argues Francisco Pereira:
“Even if all vehicles were electric and powered by non-carbon emitting energy sources, we would still need smarter solutions. Just look at congestion in cities, imposing huge costs on society as thousands of people sit idle in queues.”
Denmark is not alone regarding the desire for smarter mobility, and it is only natural that the project group is internationally oriented.
“At DTU Management, we mainly focus on algorithms that can establish where people are, and where they are likely to want to go to in the nearest future. However, even the best predictions are not really useful if you do not have in place a system able to respond to the predictions and transform them into actions. Other groups in the project are stronger on that matter, so it is fruitful to cooperate.”
Connecting with other fields
Several of the international partners and associated groups will give presentations at DTU High Tech Summit 2019. Having recently joined DTU from his previous position at MIT, Francisco Pereira will attend the event for the first time.
“I am keen to see how much we may be able to benefit from new types of contacts. Normally, we attend conferences that are focused on mobility. At DTU High Tech Summit 2019, we hope to connect also with people who are not part of the transport sector but are using Machine Learning and similar advanced tools in other fields.”
“It’s no use closing your eyes to the changes. We need to constantly keep up and try to influence the process.”
Professor Jan Madsen, DTU Compute