Adaptive longitudinal control of an autonomous vehicle with an approximate knowledge of its parameters

This paper explores the longitudinal control problem of an autonomous car in legal speed range. The goal is to develop a longitudinal controller that does not rely on vehicle identification parameters, while being capable of tracking the speed profile with comfort acceleration. A modification of a Model Reference Adaptive Control (MRAC) technique found in literature has been deeply studied and implemented, by setting the proper initial conditions for the target application. The proposed architecture is capable of controlling a vehicle whose parameters are known approximately. A CarSim-Simulink joint simulation verifies the feasibility of the proposed strategy and evaluates performances of vehicles at low and high dynamics conditions.