Piecewise Affine Relaxation of Discrete Value Functions in Learning Model Predictive Control With Application to Autonomous Racing | IEEE Journals & Magazine | IEEE Xplore