Optimal Control of Automobiles for Fuel Economy

This paper describes an application of dynamic programming to determining optimal driver control of an automobile for fuel economy. The objective function is provided by a simulator that uses vehicle performance maps derived from statistical analysis of road data collected by the authors. One dynamic program controls acceleration as a function of time subject to constraints on speed, acceleration, and distance covered. Another controls acceleration and gearshift subject to constraints on speed, acceleration, and time required to shift gears. Results are presented for acceleration to a given cruising speed, driving over hills while achieving a given average speed, and driving from one stop sign to another.