An extended intelligent driving model for autonomous and manually driven vehicles in a mixed traffic environment with consideration to roadside crossing

Abstract

While the advantages of Autonomous vehicles (AVs) and their impact on manually-driven vehicles (MVs) have been widely discussed in continuous flow conditions, their performance under mixed traffic, intermitted flow conditions has yet to be properly studied. One of the representative scenarios is that vehicular flow is interrupted by roadside crossing obstacles such as pedestrians or cyclists. Since such interruption makes vehicles stop and go more frequently and creates random and complex traffic conflict, it has become a critical factor that can affect the driving performance of AVs. Therefore, this paper proposes a uniform traffic model (Pre_IDM+) to include roadside crossing impact in traffic flow analysis. The classical intelligent driving model (IDM) is extended into an obstacle-avoiding case, in which a novel pre-reaction workflow is introduced to describe yielding behavior and generate a reasonable braking trajectory. A real mixed traffic data near an un-signalized mid-block crosswalk is used to calibrate Pre_IDM+ and an accordingly microscope mixed traffic simulation platform is constructed. The simulation results show that discreet AVs can greatly avoid hard braking (−83.61%) and slightly improve passing speed (+5.11%) compared with MVs, while competitive AVs can maximize traffic efficiency (+7.03%) but will also deteriorate driving smoothness and comfort (−31.66%). Maintaining a sparse distribution of crossing sites along the road may contribute more to traffic stability and driving continuity compared with gathering all obstacles crossing at one location. This paper may help better understand the impact of AVs on general intermitted flow and give a reference to mixed traffic modeling towards a complex road condition.