Behavioural data analysis

Topic leaders: Claudio Lantieri (UNIBO) and Rachid Belaroussi (Uni. Eiffel)

The “behavioural data analysis” topic covers the study, the analysis and the modelling of the travel behaviours of different road users, according to the characteristics related to the individual, vehicle, infrastructure and the environment. In particular, UNIBO conducts studies on pedestrian mobility, vulnerable road users, road-driver interaction, and the observation and analysis of trajectories. The goal is to train and develop behavioural knowledge in order to reduce the number and severity of road accidents. For the past five years, deep neural networks have revolutionized computer vision and speech recognition, and show promising results in other areas such as language comprehension and machine translation. However, these techniques are only starting to be implemented in the field of intelligent transport, mainly for the perception around or inside the vehicle. Uni. Eiffel scientists use these tools to visualize complex data and large dimensions of driving data (heading, speed, long / lateral acceleration, braking, steering angle but also gaze direction, EEG, EKG) collected in naturalistic driving studies or driving simulator studies. The collaboration between Uni. Eiffel and UNIBO focusses on unsupervised learning methods highlighting particular events (accidents, intersections, traffic controls, yield way) to facilitate the exploitation of data regarding road safety and user behaviour.

The synergy stands at the level of the interaction between road configuration and users’ behaviour. A Master internship has already been co-supervised, and a Cleardoc PhD is in progress, started from novembre 2023. There are perspectives to implement the Transpolis facility to collect behavioural in controlled conditions.