Investigation on the interaction between autonomous vehicles and pavement rutting considering driving safety

Autonomous vehicles (AVs) impart more potential lateral distribution modes compared to human-operated vehicles. Particularly, channelized distributions may aggravate and reshape pavement rutting, which, in turn, induces severe safety risks. Despite the critical importance of AVs' safety, a few studies have examined the interaction between AVs and pavement rutting from the perspective of driving safety. This paper presents an analytical framework to bridge this gap. We first model three AVs' wander modes and employ the finite element method to simulate rutting evolution under AV loads. Then the rutting transverse profile (RTP) is imported into CarSim. Next, the full-vehicle model and automatic controller are integrated to evaluate rollover and hydroplaning risks of AVs on the rutted road. The results indicate that more centralized load distributions yield greater peak rutting depth (RD), contrasting with trends of other RTP parameters. During crossing rutting, the zero wander mode poses the greatest rollover threat. During emergency braking in the ponding rutting, all modes lead to hydroplaning above 100 km/h. Notably, the driving safety of AVs is influenced by RTP rather than a single RD indicator. This study has potential applications for the maintenance and upgrading of road infrastructure, as well as pre-warning of AVs in terms of the operational design domain (ODD).