Perception, most specially visual perception, fundamentally determines behaviors on the road and in the street.

The road and the vehicle are equipped with devices which ensure the safety and comfort of their users by allowing them to see and to be seen at all time. The design of such equipment relies on knowledge and models of how the human visual system works (visibility, glare) and how travelers collect information in their visual environment (conspicuity, legibility). It also relies on the knowledge of the bidirectional photometric properties of the road surfaces (luminance coefficient), and how they evolve in time and weather (wet/dry).

Among the sensors implemented by these systems, cameras play a prominent role because they can often perform several functions (guidance, obstacle detection). However, their capacity to operate in poor visibility conditions (nighttime, fog, rain) still needs improvement (visibility restoration), in the visible and infrared domains. Assessing the uncertainty of the information provided by computer vision algorithms will contribute to the reliability of assistance and autonomous systems.

In a smart and cooperative transport system, the data collected by such sensors can be shared with road operators to monitor the wear of the equipment of their network. In return, roadside cameras can help vehicles by extending their field of perception and improving the accuracy of their localization.