Intelligent Vehicle Left-turn Trajectory Planning at Intersections

With the development of intelligent vehicles, trajectory planning as its key technology needs to be improved. In order to improve the real-time and anthropomorphic left-turn automatic driving behavior of intelligent vehicles at intersections, we propose a left-turn trajectory planning algorithm considering safety, energy saving, and comfort, which eliminates the influence of intersection lane number factor and its adaptability depends on the constraints of key variables such as vehicle left-turn time, longitudinal and lateral velocity, acceleration, and road boundary. A structured base scenario of a cross-type planar intersection is constructed, and the vehicle left-turn safety zone under a dynamic environment is analyzed based on vehicle kinematics and dynamics theory. Based on the Interaction dataset to extract intersection vehicle left-turn trajectories for data analysis, a fourth degrees polynomial was determined as the vehicle left-turn trajectory cluster model. The evaluation functions of vehicle left-turn safety, economy and comfort based on the feasible solution space are analyzed, a multi-performance objective co-optimization model and constraints are constructed, and a particle swarm optimization algorithm is applied to optimize them. The multi-objective left-turn trajectory planning algorithm under structured road scenarios is empirically tested by using Matlab software to further confirm the accuracy of the multi-performance objective collaborative optimization model. The simulation results demonstrate that the left-turn trajectory planning algorithm proposed can produce a smooth, energy-saving, and safe left turn when the constraints are satisfied. Additionally, the left-turn process’s safety, economy, and comfort are improved by 9.1%, 69.5%, and 7.2%, respectively, compared to the human-driven left-turn trajectory, which significantly enhances the anthropomorphism and real-time of the left-turn trajectory planning of inside.