This study is part of the SURCA French project, which goal is to identify relevant scenarios for interaction between autonomous vehicles (AV's) and other road users. Observations from current AV deployment in real traffic have raised that braking situation are a major source of conflict. The objective of this study is to collect human driven vehicle's deceleration profiles, and explore the link between their characteristics and the reason to stop. Existing studies on drivers' braking behaviour investigated data either from controlled experiments or driving simulators. Few studies have explored data from naturalistic driving behaviour, and none of them studied the link between the deceleration behaviour and the intersection characteristics. Almost 4000 deceleration profiles have been extracted from previous experiments, and analysed according to the priority context. The results indicate that significant different behaviour is observed for roundabouts, give-way, and stop-signs, with stronger decelerations and higher approach speeds.