Development of Automatic Control for a Hydraulic Excavator

Abstract

Hydraulic excavators are inherently nonlinear due to the hydraulic servo system and interaction forces that arise from soil-tool interaction. Such aspects make designing controllers for hydraulic excavators a challenging task. In this work, we develop a control framework and trajectory generation algorithm with the aim of autonomous excavation in mind. Within the proposed control framework, two controllers are embedded in a cascade manner: a proportional-derivative (PD) controller in the inner-loop and an adaptive sliding mode controller in the outer-loop. The PD controller is utilized such that the closed-loop dynamics of the inner-loop can be modeled as a linear system with parametric uncertainty. The outer-loop control is introduced to overcome the model uncertainties and external disturbances. The trajectory generation algorithm developed in this work follows two steps. First, several waypoints are generated by a proposed algorithm, which constitute a complete digging cycle. Then an optimal trajectory that satises the waypoints is obtained by means of virtual motion camou age (VMC). VMC is a biologically inspired optimization technique that reduces the solution space of the optimization problem. As a result, local optimal trajectories can be obtained with little computation. Leveling requires a higher precision of the bucket tip control when compared to digging. In order to enhance the performance of leveling, we introduce contour control to regenerate the reference trajectory for leveling. Experimental results, using a 21-ton class hydraulic excavator, of the developed control framework and algorithms for trajectory generation are presented.