CAN-based networked path-tracking control of a 4WS4WD electric vehicle

Abstract

A four-wheel steering four-wheel drive (4WS4WD) electric vehicle has a steering motor and a driving motor for each wheel, for a total of eight motors. An earlier work of the authors [2] presented a multi-input multi-output (MIMO) path-tracking control system (PTCS) for an autonomous version of this vehicle. The practical implementation of the PTCS planned by the authors has these nine modules communicating and forming feedback loops over a Controller Area Network (CAN)-based serial link, thereby forming a networked control system (NCS). However, the MIMO nonlinear loops in the PTCS turn the selection of the sampling period TS for a digital implementation, while factoring the time delays introduced by the communication, into a non-trivial task. This work solves the difficulty of MIMO nonlinear loops by finding a SISO representation of the MIMO NCS using a procedure that is applicable to a class of MIMO NCS that evince a certain symmetry. This work solves the problem of conservativeness of the controller and order of controller by systematically accounting for the time delays caused by the communication and by controller code execution. It then validates the choice of TS through a hardware-in-the-loop simulation. The techniques shown in this work are promising for applications involving the coordination of multiple actuators and for CAN-based NCS. Almost all of the subsequent literature seems to have focussed on the challenges, and has not tried to construct an NCS where these challenges may be absent. In sharp contrast, this work focuses in the positives of the NCS architecture, and avoids the challenges by using the communication protocol carefully. As a consequence, this work shows another positive of the NCS architecture that seems to have been overlooked by all the existing literature: that of performance improvement. This paper shows that distributed processing can be used to reduce the sampling interval.