Collision-free emergency planning and control methods for CAVs considering intentions of surrounding vehicles

Abstract

Autonomous emergency braking (AEB) systems are able to control vehicles as needed to avoid vehicle rear-end collisions. However, these systems are ineffective in scenarios with laterally cut-in vehicles and rapidly-changing dangerous scenes. This paper proposes a novel collision-free emergency braking system (CFEBS) that can enable intelligent connected vehicles (CAVs) to plan and execute a more conservative safety trajectory for the braking process in dangerous scenes by considering the longitudinal and lateral motion intentions of the surrounding vehicles. An intention identification model for surrounding vehicles is proposed based on long–short term memory (LSTM) networks and conditional random fields (CRFs). By considering the surrounding vehicles as risk sources and quantifying the risk with the speed of the risk flow, a potential risk flow model is built to calculate the potential risk map (PRM) around the ego vehicle. The global safest trajectory is generated via the PRM using the discrete method. The output trajectory profile is regarded as the reference for a model predictive controller (MPC). Simulation results show that the proposed CFEBS can predict vehicle intention with 91.6% accuracy and control the ego vehicle to perform effective collision-free braking operations in emergency traffic environments.