Modelling of robotic bulldozing operations for autonomous control

The low-level modeling and control of mobile robots that interact forcibly with their environment, such as excavation machinery, is a challenging problem in robotics research. This paper investigates the modeling of a robotic bulldozing operation for the purpose of autonomous control. The distinct operating modes are contained within a hybrid system modeling framework. The dynamics of the individual modes are represented by a set of five nonlinear differential equations. Model parameter estimation and validation were completed using experimental data from a scale model bulldozer. The average normalized root-mean-square prediction errors for the five states were 0.9%, 3.1% and 4.4% for one-step, five-step and ten-step ahead prediction horizons, respectively. The model will be used for the future development of model-based optimal bulldozing control.