Off-road mobile robots control: An accurate and stable adaptive approach

Abstract

Automation in outdoor applications at high speed (such as farming, surveillance, etc.) requires the highly accurate control of mobile robots, taking into account natural ground characteristics, in order to preserve both motion accuracy and robot stability. To meet such expectations, advanced control laws must be developed. In previous studies, adaptive and predictive control algorithms, based on both an extended kinematic and a dynamic representation, have been proposed to specifically address path tracking for such robots. However, even if high accuracy path tracking can be attained, such control laws do not enable roll-over situations to be avoided. This paper therefore addresses the prevention of off-road mobile robot roll-over as well as path tracking. In order to meet this aim, a stability metric, depending on a dynamic model, is computed and sliding phenomena are accounted for via a mixed observer (a multi-model observer, mixing kinematic and dynamic modelling). The steering angle is then controlled by the path tracking algorithm, and robot longitudinal speed may be limited to ensure safe motion. More precisely, the robot velocity leading to a critical value of the stability metric is computed via Predictive Functional Control (PFC), and this constitutes the speed control set point when inferior to the desired robot speed. The capabilities of the proposed approach are demonstrated and discussed thanks to full scale experimentation.