System Identification and Adaptive Steering of Tractors Utilizing Differential Global Positioning System

A real-time gradient-based method of identifying parameters deŽ ning the dynamic response of a farm tractor is presented. By use of a linear state-space model of the vehicle, four parameters are identiŽ ed with carrier-phase differential Global Positioning System as the primary sensor for position and attitude, and a linear potentiometer is used to measure steering angle. The resulting real-time estimates of the vehicle parameters are used to redesign linear control gains. This adaptive estimation and control scheme achieves a 25% improvement in lateral error standard deviation over a conventional Ž xed-gain controller.