Recovering Models of a Four-Wheel Vehicle Using Vehicular System Data

This paper discusses efforts to parameterize the actuation models of a four-wheel automobile for the purposes of closed-loop control. As a novelty, the authors used the equipment already available or in use by the vehicle, rather than expensive equipment used solely for the purpose of system identification. After rudimentary measurements were taken of wheelbase, axle width, etc., the vehicle was driven and data were captured using a controller area network (CAN) interface. Based on this captured data, we were able to estimate the feasibility of certain closed-loop controllers, and the models they assumed (i.e., linear, or nonlinear) for control. Examples were acceleration and steering. This work served to inform the separation of differences in simulation and vehicle behavior during vehicle testing. I. INTRODUCTION A major complexity of vehicle control is an accurate model of the vehicle for controller design, and simula- tion. Poor vehicle models can result in unstable behavior when applied to hardware, resulting in unsafe situations for those involved, or frustrating demonstrations that diverge significantly from simulation results. As a result, significant hardware-in-the-loop testing is necessary prior to control system design, to minimize the risk. This is undesirable from many aspects. First, hardware-in- the-loop experiments are costly in terms of personnel time and equipment, as well as any facilities which must be rented. Additionally, it predicates the implementation of algorithms for higher-level performance on already known or well- understood platforms upon which those algorithms will run.