Pseudo Huber Functions-Based NMPC for Tracking of Agricultural Harvester

The nonlinear model predictive control (NMPC) based on pseudo-Huber functions is studied for the trajectory tracking problem of agricultural harvesters. For the unstable value of the huber function, an improved pseudo-Huber functions is designed to get better stability and control accuracy. The pseudo-Huber function, and the quadratic punishment function are combine to construct cost function. The trigonometric function of heading error is introduced to avoid the problem of heading circling. The dynamic model of the harvester is raised, and the nonlinear solver CasADi is used to optimize the solution of the controller. Subsequently, simulation is carried out in the straight line and the steering trajectory, and a new evaluation index is proposed to prove the effectiveness of the controller. In comparison with the traditional NMPC based on error cost function, the tracking accuracy is obviously improved.