MODERATED REINFORCEMENT LEARNING OF ACTIVE AND SEMIACTIVE VEHICLE SUSPENSION CONTROL LAWS

This paper is concerned with the application of reinforcement learning to the dynamic ride control of an active vehicle suspension system. T he study makes key extensions to earlier simulation work to enable on- line implementation of the learning automaton methodology using an act ual vehicle. Extensions to the methodology allow safe and continuous l earning to take place on the road, using a limited instrumentation set . An important new feature is the use of a moderator to set physical l imits on the vehicle states. It is shown that the addition of the mode rator has little direct effect on the system's ability to learn, and a llows learning to take place continuously even when there are unstable controllers present. The study concludes with the results of an exper imental trial using vehicle hardware, where the successful synthesis o f a semi-active ride controller is demonstrated.