Optimal lane-changing trajectory planning for autonomous vehicles considering energy consumption

The emerging autonomous vehicles (AVs) technologies provide a new opportunity to design a better lanechanging trajectory to reduce traffic congestion. The existing research on lane-changing trajectory planning mainly focuses on fitting trajectory curves, and the lane-changing trajectory is not optimized from the perspective of optimal control. To address this limitation, this paper proposes an optimal lane-changing trajectory planning model for AVs. First, the vehicle lane-changing process is modeled as an optimal control problem to minimize fuel consumption. The control variables are acceleration and steering angle. Second, to simplify the optimal control model, the speed and acceleration are decomposed in X and Y directions. The nonlinear constraints of the optimal control model are transformed into linear constraints. Then, we discretize the time of the optimal control model and simplify the problem into nonlinear programming (NLP). Finally, taking the NGSIM database and polynomial trajectory curve as the benchmark, we verify that the optimal lanechanging trajectory can significantly save energy consumption, with an average reduction of 37.52% and 32.48%. Sensitivity analysis indicates that (1) the minimum fuel consumption of lane-changing trajectory corresponds to optimal lane-changing duration and an optimal longitudinal displacement; (2) the more significant the speed difference (i.e., initial speed and final speed) before and after lane-changing, the greater the fuel consumption of lane-changing; (3) the fuel consumption of lane-changing increases with the increase of lateral displacement.