Provably Safe and Smooth Lane Changes in Mixed Trafic

Abstract

While lane change behavior of human drivers has already been widely investigated, concepts and algorithms to plan lane changes for automated vehicles become necessary through the progress of automated driving. In this paper, we present an approach for safe and smooth lane changes in dense traffic, that consider both the cooperation of humans but also the increased need for safety compared to lane changes performed by human drivers. This real time capable approach is based on path-velocity decomposition, the intelligent driver model, a model compliance check for human machine cooperation and the concept of responsibility sensitive safety. The work is targeted towards mandatory lane changes determined by an upstream routing module, but is also applicable to optional lane changes.