On state and inertial parameter estimation of free-falling planar rigid bodies subject to unscheduled frictional impacts. (November 2019)

• Record Type: Journal Article
• Authors: Gabiccini, Marco; Fusco, Francesco
• Is Part Of: Mechanism and machine theory. Volume 141(2019)
• Page Start: 171
• Page End: 195
• Publication Date: 2019-11
• Subjects: Inertia estimation -- Contact dynamics -- Frictional impacts -- Nonlinear optimization; Machine theory -- Periodicals; Machinery -- Periodicals; Machines -- Périodiques; Génie mécanique -- Périodiques; Machine theory; Machinery; Periodicals; 621.81
• DOI: 10.1016/j.mechmachtheory.2019.07.010
• Languages: English
• ISSNs: 0094-114X

Abstract

Highlights: Estimation of states, friction and inertia parameters of a rigid body from a video. Rigid body falls under gravity and impacts the ground in an unscheduled contact sequence Problem framed as a large scale nonlinear optimization and solved efficiently. Inertia-to-weight ratio identification accuracy increases for higher friction. Inertia-to-weight ratio identification accuracy increases for higher camera frame rate. Saturation on the estimate of friction coefficient when motion becomes pure rolling. Abstract: This paper addresses the problem of simultaneous state estimation and inertial and frictional parameter identification for planar rigid-bodies subject to unscheduled frictional impacts. The aim is to evaluate to what level of accuracy, given noisy captured poses of an object free-falling under gravity and impacting the surrounding environment, it is conceivable to reconstruct its states, the sequence of normal and tangential impulses and, concurrently, estimate its inertial properties along with Coulomb's coefficient of friction at contacts. To this aim we set up a constrained nonlinear optimization problem, where the unscheduled impacts are handled via a complementarity formulation. To assess the validity of the proposed approach we test the identification results both (i) with respect to ground truth values produced with a simulator, and (ii) with respect to real experimental data. In both cases, we are able to provide accurate/realistic estimates of the inertia-to-mass ratio and friction coefficient along with a satisfactory reconstruction of systems states and contact impulses.