The objective of this work is to optimize control strategy of a vehicle which leads to low fuel consumption. A specialized car developed especially to take part in the Shell Eco-marathon competition is considered. At first a general first principle model of the vehicle is discussed, its parameters are tuned taking into account the data obtained experimentally and the model is validated using real measurements. Next, the optimization problem leading to low fuel consumption is discussed. The engine usage continuous-time control strategy is parametrized by means of 4 methods in order to reduce computational complexity. Since the resulting constrained optimization problem is nonlinear, as many as 5 evolutionary algorithms are examined in terms of quality of the final result and computational efficiency. Inefficiency of a classical gradient-based optimization method is demonstrated. Finally, a complete decision system which can act as an autonomous vehicle engine controller is discussed.