Design and Comparison of Fuel-Saving Speed Planning Algorithms for Automated Vehicles

Intelligent planning and accurate execution of connected automated vehicles (CAVs) enable not only improved traffic safety but also better fuel economy. This paper presents two longitudinal speed planning algorithms for fuel-saving driving on highways with varying road slopes. One is designed on the top of the model predictive control (MPC) and the other is called equivalent kinetic-energy and fuel conversion method. The MPC algorithm solves the optimal speed profile in a receding finite horizon with repeated optimization, which is numerically solved by the Legendre pseudospectral method. The latter is designed based on an instantaneous optimization, which considers vehicle kinetic energy an admissible power source, and then minimizes a weighted sum of fuel energy and kinetic energy. This strategy is capable of generating analytical rules to get the economical speed as well as the corresponding commands of the engine, transmission, and brake. The two algorithms are featured by near-global optimization and local optimization, respectively. Their performances in fuel economy and computational load are quantitatively explored and compared in order to distinguish the potential of real implementation in CAVs.