Application of automatic differentiation to race car performance optimisation

Abstract

A formal method for the evaluation of the minimum time vehicle manoeuvre is described. The problem is treated as one of optimal control and is solved using a direct transcription method. The resulting nonlinear programming problem is solved using the sequential quadratic programming algorithm SNOPT for constrained optimisation. The automatic differentiation software too ADoptis used for the evaluation of the first-order derivatives of objective and constraint functions with respect to the control variables. The implementation of automatic differentiation is more robust and ten times as fast compared to the use of a finite difference determination of the Jacobian.