A nonsmooth nonlinear conjugate gradient method for interactive contact force problems

Morten Silcowitz; Sarah Maria Niebe Abel; Kenny Erleben

doi:10.1007/s00371-010-0502-6

A nonsmooth nonlinear conjugate gradient method for interactive contact force problems

Morten Silcowitz, Sarah Maria Niebe Abel, Kenny Erleben

Department of Computer Science

15 Citations (Scopus)

Abstract

Interactive rigid body simulation is important for robot simulation and virtual design. A vital part of the simulation is the computation of contact forces. This paper addresses the contact force problem, as used in interactive simulation. The contact force problem can be formulated in the form of a nonlinear complementarity problem (NCP), which can be solved using an iterative splitting method, such as the projected Gauss–Seidel (PGS) method. We present a novel method for solving the NCP problem by applying a Fletcher–Reeves type nonlinear nonsmooth conjugate gradient (NNCG) type method. We analyze and present experimental convergence behavior and properties of the new method. Our results show that the NNCG method has at least the same convergence rate as PGS, and in many cases better.

Original language	English
Journal	The Visual Computer
Volume	26
Issue number	6
Pages (from-to)	893-901
Number of pages	9
ISSN	0178-2789
DOIs	https://doi.org/10.1007/s00371-010-0502-6
Publication status	Published - Jun 2010

Access to Document

10.1007/s00371-010-0502-6

Cite this

@inproceedings{05c726b8e9d645efb87a667b352a5b84,

title = "A nonsmooth nonlinear conjugate gradient method for interactive contact force problems",

abstract = "Interactive rigid body simulation is important for robot simulation and virtual design. A vital part of the simulation is the computation of contact forces. This paper addresses the contact force problem, as used in interactive simulation. The contact force problem can be formulated in the form of a nonlinear complementarity problem (NCP), which can be solved using an iterative splitting method, such as the projected Gauss–Seidel (PGS) method. We present a novel method for solving the NCP problem by applying a Fletcher–Reeves type nonlinear nonsmooth conjugate gradient (NNCG) type method. We analyze and present experimental convergence behavior and properties of the new method. Our results show that the NNCG method has at least the same convergence rate as PGS, and in many cases better. ",

author = "Morten Silcowitz and Abel, {Sarah Maria Niebe} and Kenny Erleben",

year = "2010",

month = jun,

doi = "10.1007/s00371-010-0502-6",

language = "English",

volume = "26",

pages = "893--901",

journal = "Visual Computer",

issn = "0178-2789",

publisher = "Springer",

number = "6",

}

TY - GEN

T1 - A nonsmooth nonlinear conjugate gradient method for interactive contact force problems

AU - Silcowitz, Morten

AU - Abel, Sarah Maria Niebe

AU - Erleben, Kenny

PY - 2010/6

Y1 - 2010/6

N2 - Interactive rigid body simulation is important for robot simulation and virtual design. A vital part of the simulation is the computation of contact forces. This paper addresses the contact force problem, as used in interactive simulation. The contact force problem can be formulated in the form of a nonlinear complementarity problem (NCP), which can be solved using an iterative splitting method, such as the projected Gauss–Seidel (PGS) method. We present a novel method for solving the NCP problem by applying a Fletcher–Reeves type nonlinear nonsmooth conjugate gradient (NNCG) type method. We analyze and present experimental convergence behavior and properties of the new method. Our results show that the NNCG method has at least the same convergence rate as PGS, and in many cases better.

AB - Interactive rigid body simulation is important for robot simulation and virtual design. A vital part of the simulation is the computation of contact forces. This paper addresses the contact force problem, as used in interactive simulation. The contact force problem can be formulated in the form of a nonlinear complementarity problem (NCP), which can be solved using an iterative splitting method, such as the projected Gauss–Seidel (PGS) method. We present a novel method for solving the NCP problem by applying a Fletcher–Reeves type nonlinear nonsmooth conjugate gradient (NNCG) type method. We analyze and present experimental convergence behavior and properties of the new method. Our results show that the NNCG method has at least the same convergence rate as PGS, and in many cases better.

U2 - 10.1007/s00371-010-0502-6

DO - 10.1007/s00371-010-0502-6

M3 - Conference article

SN - 0178-2789

VL - 26

SP - 893

EP - 901

JO - Visual Computer

JF - Visual Computer

IS - 6

ER -