How can traffic flows be simulated and modeled and how can we develop new technologies for an efficient, safe and sustainable transport system of the future?

Mathematical and statistical methods for the analysis, modelling and optimization of traffic flows are researched at the Chair of Econometrics and Statistics, esp. in the Transport Sector. Using empirical data and testing new methodological approaches, developments in the transport sector can be forecast.

In the Reinforcement Learning (RL) research group founded by Prof. Okhrin, scientists are working on reinforcement learning as a form of machine learning. In contrast to supervised learning, in reinforcement learning a software agent learns through long-term rewards, whereby it behaves independently in a dynamic environment.

“We train our vehicles so that they learn from themselves, just like a baby. We don't tell it how to behave, but we set the goal. If it leaves the lane on its way, it is sanctioned. If it has a collision, it is sanctioned. That's how it does tens of millions of trials in our simulation environment,” says Prof. Okhrin, describing the procedure.

© Fakultät Verkehrswissenschaften

The potential of reinforcement learning lies in the automation of error- and accident-prone tasks. It is therefore particularly suitable for unsafe and complex scenarios such as autonomous driving and can be applied to a real traffic environment to improve the control of traffic dynamics:

“The most exciting thing about it is how we get from simulation to reality. We are working very actively on how we can close this gap, the so-called Sim2Real Gap,” says Prof. Okhrin.

Prof. Ostap Okhrin is delighted with the award: “Yes, it was also an extra pleasure for me that I managed to be listed there. Especially as I have published more in the fields of AI and automated traffic in the last two years.”

We congratulate Prof. Okhrin on his inclusion in the Top 100 Young Economists.

By the way: Prof. Okhrin is actively involved in science communication. During the Dresden Long Night of Science, he presents the projects and winners of the Nobel Prize in Economic Sciences live. He also creates explanatory videos afterwards, in which the topics are explored in greater depth and presented to a wider audience. To the YouTube Playlist