Find in Library

This study proposes a trajectory planning method for an automated lane change maneuver. We consider that the spatiotemporal trajectory of a controlled vehicle can be represented by a polynomial function and estimate the parameters of the polynomial function through nonlinear programming that does not rely on the assumed time horizon of lane change and the assumed final state of the controlled vehicle. When the controlled vehicle achieves its target lane, the state of this vehicle should be constrained by both the position and the speed of the forward vehicle. We integrate a car-following model into the nonlinear programming to constrain the state of the controlled vehicle at the end of the lane change. Notably, the interaction factors are taken into consideration: the motion of the vehicle that follows the controlled vehicle in the target lane can be influenced by the trajectory planning results of the controlled vehicle. The proposed trajectory planning method can simultaneously estimate the motion of the following vehicle and plan the trajectory for the controlled vehicle. Our proposed model can also work for curved road sections. We represent the curve information in the nonlinear programming by a regression model. We solve the nonlinear programming through the sequential quadratic program (SQP) algorithm and design a method to specify an initial guess for the algorithm. This paper finally demonstrates the effect of the proposed trajectory planning method under different scenarios.