Lane Change Behavior Modeling for Autonomous Vehicles Based on Surroundings Recognition

This paper describes an algorithm that determines the reference vehicle trajectory for autonomous lane changing maneuver. The algorithm employs potentially acquired on-board vehicle sensing signals like the velocity of each vehicle, the relative distance between vehicles as variables and estimates the positions of host and surrounding vehicles after finishing lane change maneuver. The lane change timing and reference lateral/longitudinal acceleration are derived theoretically and compared with the data which is obtained from human driver experiments carried out on driving simulator. The longitudinal and lateral accelerations are determined as the reference model. The paper describes algorithm details, following by simulation results which show the feasibility of the proposed algorithm.