A solution to apply load-transfer including vehicle model to optimal control formulation

Abstract

The aim of this paper is to build a methodology to predict the evaluation of vehicle's dynamic performance. At the evaluation process, driver is integrated into driver-vehicle system. Therefore, if we estimate the evaluation grades, a specific mechanical property of vehicle is not sufficient but multilateral properties related to driver's characteristics are required. On experiment side, there are two way: measuring vehicle responses related to driver feeling (open-loop approach) and marking achievement level of driving task related to driver's driving intention (closed-loop approach). On analytical side, open-loop approach is easy to execute with detailed vehicle model, closed-loop requires optimal control methods which demands strong restriction to model complexity. The primary reason for model limitation comes from incompatible formulation in jacking force, a part of load transfer expression among each wheel. To settle this problem, this paper investigated the problematic part of jacking force and inserted virtual mechanism to convert tire lateral forces into load transfer with jacking force. Along with this formulation, verification procedure to check the tolerance of load transfer disparity is built. When optimal control calculated vehicle motion, path following vehicle model with correct jacking force expression identify the accuracy of computed result. As a result of these procedural steps, practical level vehicle model, including tire load transfer mechanism, is now applicable to optimal control problem. For exemplify the usefulness of this methodology, simulation of lane change task with different load transfer property (roll stiffness distribution) is shown. When we use driver modeling for closed-loop simulation based on forward dynamics, there always exists ambiguity of the modeling. With optimal control simulation, we can specify driver's driving intention with terminal constraints and cost function more clearly. This paper's methodology expands the possibility to estimate the evaluation of vehicle dynamic performance on closed-loop simulation.