A novel adaptive fuzzy reinforcement learning controller for a platoon of off-axle hitching tractor-trailers with a prescribed performance and path curvature compensation

Abstract

This paper addresses an artificial intelligence-based platoon control (AIPC) problem of autonomous off-axle hitching tractor-trailer wheeled mobile robots (TTWMRs) in a planar motion. Towards this end, a novel saturated nonlinear interval type-2 neuro-fuzzy reinforcement learning controller is proposed to construct a convoy-like motion of TTWMRs. To overcome the cutting-corner phenomenon in the platoon motion, the curvature of the reference trajectory is estimated and the cutting-corner behaviour is prevented. In order to deal with the situation where the velocity and acceleration signals are not measurable or corrupted by the measurement noise, a high-gain observer (HGO) is employed. The prescribed performance control (PPC) is also engaged in the control design procedure to guarantee inter-vehicular communication preservation, collision prevention between each consecutive pair in the platoon, singularity cancellation, and some prespecified desired characteristics of the convoy formation tracking errors such as maximum undershoot/overshoot, convergence rate, and final tracking precision. The Lyapunov-based stability analysis reveals semi-global uniform ultimate boundedness (SGUUB) convergence of tracking errors for the entire platoon system. With the suggested control framework, simulations and comparisons are conducted in which the control performance is considerably improved compared with the existing literature.