Driving safety field modeling focused on heterogeneous traffic flows and cooperative control strategy in highway merging zone

Abstract

Connected and Automated Vehicles (CAVs) will coexist with Human-Driven Vehicles (HDVs) in the foreseeable future. Proactive safe controls indeed play a vital role in thus a complex heterogeneous traffic environment. This paper mainly constructed the safety field models for vehicles and environments respectively to address the problems of unclear rules of heterogeneous traffic flow and unclear safety mechanisms in the complex scenes of merging areas. Firstly, the action range of CAV vehicle field was constrained with the differential approximation algorithm to fill in the weakness of the infinite influence range of the existed vehicle field. Then the safety field model of HDV was also reconstructed by taking into account the driver’s environmental psychological tolerance with three indicators from driver experience, environmental visibility and trust. Meanwhile, the environmental field model was innovatively set up with the longitudinal distance variable to overcome the drawbacks of sudden changes in field strength as the number of lanes changes. Based on above mechanism models, this study further reconstructed the early lane-changing model that cored with the minimum distance model and the mandatory lane-changing model in the merging zone that cored with the driving safety index (DSI). Those models provided explicit judgment for the selection of vehicle lane-changing model and CAV behavior control in the merging zone. Finally, the simulation experiments show that the cooperative control strategy with a higher of 45% CAV penetration rate in low-density traffic (1200 veh/h) scenario outperforms than those strategies of SUMO or without early lane-changing in adapting the features of highway merge zone. While in a high-density traffic (1600 veh/h) scenario that with a fixed 30% CAV penetration rate, the time-average speed and space-average speed of the cooperative control strategy improved 15.56% and 25.15%, respectively.