Unknown-input observer design for motorcycle lateral dynamics: TS approach

Abstract

In this paper, a nonlinear observer is designed in order to estimate the lateral dynamics of motorcycles. A nonlinear model of motorcycle's lateral dynamics is considered and is transformed in a Takagi-Sugeno (TS) exact form. An unknown input (UI) nonlinear observer is then designed in order to reconstruct the state variables whatever the forward velocity variations. The observer convergence study is based on the Lyapunov theory. The boundedness of the state estimation error is guaranteed thanks to the Input to State Stability (ISS) property. The observer has been tested on a nonlinear multibody model.

Conservatism reduction for Nonlinear Takagi-Sugeno Observer : Interconnected System Approach

“…To rewrite the previous example (8) into the classical Takagi-Sugeno representation it is necessary to invert the matrix E(ν) resulting in (9). This gives r = 2 3 = 8 sub-models.…”

“…In [6], [7], [8], observers have been proposed for state estimation and state estimation in the presence of unknown inputs (faults, perturbations ...). Extensions have been proposed for state estimation of Takagi-Sugeno systems with state dependent premise variables [9], [10], [11] and differentiation techniques [12]. Notice that all these works transform the nonlinear system in a TS form and design the observer *This work was not supported by any organization The author is with Evry Val d'Essonne University, IBISC-Lab, 40, rue de Pelvoux, 91020, Evry Courcouronnes, France (email: {hichem.arioui}@univ-evry.fr) which leads, in several cases of strong nonlinear systems, to TS systems with a huge number of sub-models and then a huge number of LMIs to solve.…”

Rider weight consideration for observer design with an application to the estimation of the lateral motorcycle dynamics and rider's action

“…Previous studies have ever deal with the rider's weight uncertainty like in [15] where the author studies the influence of the rider's properties on the PTWV stability especially on the weave mode. In [16] or [17], authors used a common method to include rider mass uncertainty which is taken into consideration as an uncertain state matrix and input vector such as in the state space representation A = A nominal + ∆A and B = B nominal + ∆B, with ∆A and ∆B containing the uncertainties regarding the nominal case. Then, the robust techniques such as L 2 -gain are used to minimize the transfer of the term due to the uncertainties on the estimation errors.…”

Estimation of lateral motorcycle dynamics and rider action with Luenberger observer

“…Longitudinal dynamic estimation of PTW is discussed in [8] or [9], whereas lateral estimation is addressed in [5], [10], [11], [12], [13] and recently in [14]. Concerning lateral estimation most of cited papers consider restrictive assumptions which regarding riding motorcycle practices are not in adequation with the reality.…”