Research topics

PICS-L's ambition is to produce knowledge and tools to improve the safety and comfort of individual travel while preserving the environment. The challenge is to support the evolution of mobility in the context of ecological, energy, and digital transitions. We are particularly interested in motorists, active modes of transport, and automated vehicles, which we consider to be future users.

The research conducted by PICS-L is divided into three cross-disciplinary topics covering different aspects of the perception-action loop that underpins locomotion: perception, behavior and action, and immersive simulation. Our work has two dimensions: between humans and robots on the one hand, and between real and simulated travel environments on the other.

• In the topic of perception, we are interested in how information acquisition systems work, especially visual ones, and in the phenomena that disrupt this functioning in difficult situations, particularly due to unfavorable lighting and visibility conditions. We deal with human vision and computer vision, the common factor being the visual “signal” from which useful information must be extracted. In both cases, we seek to characterize and improve visibility in the driving environment, using “real” images collected in the laboratory, on the track or on the open road, as well as images produced by digital simulation.

• In the topic of behavior and action, we focus on interactions between users, their vehicles, infrastructure, and other users in scenarios that pose challenges for safety, operations, or the environment. We examine human behavior, particularly its psychological determinants, as well as system dynamics, with a view to controlling them.

• In the topic of immersive simulation, we are interested in the accuracy of virtual reality and motion simulation devices that enable us to study perception and behavior under controlled and repeatable conditions. We deal with humans in the loop, but also software in the loop, because while simulation is essential for studying behavior under controlled and repeatable conditions, it is equally essential for evaluating the reliability of automated systems, particularly when they use artificial intelligence. To ensure that the behaviors observed in simulation are consistent with those that would be observed in real-life situations, real-simulated comparisons are made as much as possible. 

_{Translated with DeepL.com (free version)}