Double Gradient Method: A New Optimization Method for the Trajectory Optimization Problem

In this paper, a new optimization method for the trajectory optimization problem is presented. This new method allows to predict racing lines described by cubic splines (problems solved in most cases by stochastic methods) in times like deterministic methods. The proposed Double Gradient Method (DGM) is not affected by the dimensionality of the problem. Comparison of the results with data collected from professional drivers has shown that the DGM is reliable for lap time simulations with race line optimization. It can help drivers find the fastest racing line, be used for embedded algorithm development or for autonomous vehicle competitions.