Adaptive Observer for Motorcycle State Estimation and Tire Cornering Stiffness Identification

Abstract

In this paper, a linear parameter varying (LPV) adaptive observer is designed for state estimation and tire cornering stiffness identification based on lateral motorcycle model. The estimation is based on a general Lipstchitz condition, Lyapunov function and is subjected to persistency of excitation conditions. Further, the LPV observer is transformed into Takagi-Sugeno (T-S) fuzzy observer and sufficient conditions, for the existence of the estimator, are given in terms of linear matrix inequalities (LMIs). This method is designed assuming that some of the states are not available, since parametric identification is generally developed assuming that all the system states are available (measured or estimated). Finally, the effectiveness of the proposed estimation method is illustrated through test scenarios performed with the wellknown motorcycle simulator "BikeSim" and by field test using data measurement carried out on experimental motorcycle.

Quasi-LPV Unknown Input Observer with Nonlinear Outputs: Application to Motorcycles

“…where τ is the unknown rider's torque and v x (t) is the timevarying forward speed. Matrices Ē = [e i j ], Ā(v x ) = [a i j ] and B and their parameters are given in appendix [6].…”

“…Lot of recent works deal with motorcycle dynamic state estimation [2], [3], [4] but few of them perform experimental investigations to validate the results. In [5] and [6] experimental tests are performed but only lateral dynamics estimation is considered. In [7] authors proposed a more complete experimental investigation to validate estimated states with extended Kalman filter but only constant forward speed case is considered.…”

“…This induces significant errors in certain riding conditions (cornering, high speed, etc.). We introduced recently an interesting extension to our previous works on the estimation of the motorcycle's lateral and steering dynamics with nonlinear outputs [6] where we present here some updates as well as an experimental application to motorcycles.…”

“…The main contribution of this paper is the robustness study towards the parametric uncertainties and sensors noise and, on the other hand, the experimental validation of the proposed works in [6] by designing a new unknown input observer (UIO) with variable output matrix. The observer design is based on a simple two bodies model of the electric scooter which allows to get very good estimation time performance.…”