METHOD FOR POLYNOMIAL BASED PREDICTIONS OF EGO VEHICLES AND ADVERSARIAL AGENTS

A method for a driver prediction system is described. The method includes training a neural network to a learn a set of polynomial basis functions. The method also includes selecting, using a trained neural network, a learned polynomial basis function to perform a prediction of an action of an autonomous dynamic object (ADO) agent. The method further includes computing projection coefficients of the learned polynomial basis function to weigh the learned polynomial basis function. The method also includes using the projection coefficients of the learned polynomial basis function to weigh the learned polynomial basis function to provide a distribution regarding a likelihood of the prediction of the action of the ADO agent.