Path-Following Control for Unmanned Rollers: A Composite Disturbance Rejection-based Framework

Abstract

The drum roller, as a widely used engineering vehicle, has higher degree of freedom in motion relative to conventional passenger vehicles. The special operating condition that has large rocks on road for compaction introduces severe disturbances in path-following. In this paper, a composite disturbance rejection-based framework, for the path-following control of rollers, is proposed. The external disturbances caused by rocks on road are rejected by correcting the coordinates of rollers from Global Position System (GPS) using measured attitude information. The nonlinearities from the complex articulation structure are compensated using a kinematic model-based feedforward control. All other uncertainties, internal and external, are lumped as an augmented state - "total disturbance", estimated hence rejected in real-time via the extended state observer (ESO). As compliment to ESO with the limited performance due to low sampling rate of GPS, a model parameters self-learning algorithm is added. The proposed solution is validated both in simulation and experiments, showing satisfactory performance. The maximum lateral error is ~0.1m for unmanned rollers, out-performing the average level of human driven rollers when working on road with maximum diameter of rocks up to 1m.