Online Estimation of Implement Dynamics for Adaptive Steering Control of Farm Tractors

This paper presents a method to utilize a cascaded extended Kalman filter to estimate the slope of the dc gain M_dc of the steer angle to yaw rate transfer function with respect to the velocity of the tractor. The online estimate of the dc gain for different implements is used to compensate for changes in the yaw dynamics of the tractor implement cause by variations in the hitch loading. This is accomplished by using the online estimate of M_dc to modify the tractor controller gains based on a lookup table of predetermined values that are calculated from system identification tests for different implements. An online analysis of the data is used to determine if enough excitation in the yaw dynamics is available for estimation of the dc gain. The sensitivity of the controller to model simplifications, incorrect velocities, and M_dc estimate errors as well as the accuracy and rate of convergence of the estimated M_dc are also investigated. Experimental and simulation results are provided that show the validity of the M_dc estimate. Finally, experimental results show that the control performance is maintained in the presence of variation in hitch loading and tractor velocity using the adaptive control strategy based on the dc gain estimate for a given implementcondition.