Optimal motion planning of redundant manipulators with controlled task infeasibility

Joo Hyun Kim, Chang B. Joo

Research output: Contribution to journalArticle

Abstract

In optimal motion planning and control, the complex time-varying nature of redundant robots, environments, and task requirements causes complex domains and conflicting constraints. Since predicting or recovering infeasibility is not always possible, infeasibilities occur frequently and are not completely avoidable. We introduce a constrained nonlinear programming framework of controlled (as opposed to recovered) infeasibility for physically valid solutions while preserving the original problem and variable space. The constraint prioritization hierarchy includes a comprehensive classification of physical consistency, design requirements, and tasks. Priority weight functions having features of normalization and prioritization are incorporated into a sequential quadratic programming (SQP) algorithm to ensure generality and strict satisfaction of high-priority constraints, while lower-priority constraint violations are minimized. These are embedded in SQP through its merit function and composite cost function, in which general nonlinear functions including unilateral, time-dependent, and nonholonomic, can be incorporated in a unified approach. Also, the avoidance of the discontinuity problem with unilateral constraints is due to the time-dependent constraints strategy, which actively adapts to varying states. Numerical examples using multibody dynamic models of a redundant manipulator demonstrate these advantages.

Original languageEnglish (US)
Pages (from-to)155-174
Number of pages20
JournalMechanism and Machine Theory
Volume64
DOIs
StatePublished - 2013

Fingerprint

Redundant manipulators
Motion planning
Quadratic programming
Nonlinear programming
Motion control
Cost functions
Dynamic models
Robots
Composite materials

Keywords

  • Constraint prioritization
  • Controlled infeasibility
  • Normalization
  • Optimal motion planning
  • Redundant robot
  • Sequential quadratic programming

ASJC Scopus subject areas

  • Mechanical Engineering
  • Mechanics of Materials
  • Computer Science Applications
  • Bioengineering

Cite this

Optimal motion planning of redundant manipulators with controlled task infeasibility. / Kim, Joo Hyun; Joo, Chang B.

In: Mechanism and Machine Theory, Vol. 64, 2013, p. 155-174.

Research output: Contribution to journalArticle

@article{e05e30b32e5945b18b9a92b3b8beedbb,
title = "Optimal motion planning of redundant manipulators with controlled task infeasibility",
abstract = "In optimal motion planning and control, the complex time-varying nature of redundant robots, environments, and task requirements causes complex domains and conflicting constraints. Since predicting or recovering infeasibility is not always possible, infeasibilities occur frequently and are not completely avoidable. We introduce a constrained nonlinear programming framework of controlled (as opposed to recovered) infeasibility for physically valid solutions while preserving the original problem and variable space. The constraint prioritization hierarchy includes a comprehensive classification of physical consistency, design requirements, and tasks. Priority weight functions having features of normalization and prioritization are incorporated into a sequential quadratic programming (SQP) algorithm to ensure generality and strict satisfaction of high-priority constraints, while lower-priority constraint violations are minimized. These are embedded in SQP through its merit function and composite cost function, in which general nonlinear functions including unilateral, time-dependent, and nonholonomic, can be incorporated in a unified approach. Also, the avoidance of the discontinuity problem with unilateral constraints is due to the time-dependent constraints strategy, which actively adapts to varying states. Numerical examples using multibody dynamic models of a redundant manipulator demonstrate these advantages.",
keywords = "Constraint prioritization, Controlled infeasibility, Normalization, Optimal motion planning, Redundant robot, Sequential quadratic programming",
author = "Kim, {Joo Hyun} and Joo, {Chang B.}",
year = "2013",
doi = "10.1016/j.mechmachtheory.2013.02.001",
language = "English (US)",
volume = "64",
pages = "155--174",
journal = "Mechanism and Machine Theory",
issn = "0374-1052",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Optimal motion planning of redundant manipulators with controlled task infeasibility

AU - Kim, Joo Hyun

AU - Joo, Chang B.

PY - 2013

Y1 - 2013

N2 - In optimal motion planning and control, the complex time-varying nature of redundant robots, environments, and task requirements causes complex domains and conflicting constraints. Since predicting or recovering infeasibility is not always possible, infeasibilities occur frequently and are not completely avoidable. We introduce a constrained nonlinear programming framework of controlled (as opposed to recovered) infeasibility for physically valid solutions while preserving the original problem and variable space. The constraint prioritization hierarchy includes a comprehensive classification of physical consistency, design requirements, and tasks. Priority weight functions having features of normalization and prioritization are incorporated into a sequential quadratic programming (SQP) algorithm to ensure generality and strict satisfaction of high-priority constraints, while lower-priority constraint violations are minimized. These are embedded in SQP through its merit function and composite cost function, in which general nonlinear functions including unilateral, time-dependent, and nonholonomic, can be incorporated in a unified approach. Also, the avoidance of the discontinuity problem with unilateral constraints is due to the time-dependent constraints strategy, which actively adapts to varying states. Numerical examples using multibody dynamic models of a redundant manipulator demonstrate these advantages.

AB - In optimal motion planning and control, the complex time-varying nature of redundant robots, environments, and task requirements causes complex domains and conflicting constraints. Since predicting or recovering infeasibility is not always possible, infeasibilities occur frequently and are not completely avoidable. We introduce a constrained nonlinear programming framework of controlled (as opposed to recovered) infeasibility for physically valid solutions while preserving the original problem and variable space. The constraint prioritization hierarchy includes a comprehensive classification of physical consistency, design requirements, and tasks. Priority weight functions having features of normalization and prioritization are incorporated into a sequential quadratic programming (SQP) algorithm to ensure generality and strict satisfaction of high-priority constraints, while lower-priority constraint violations are minimized. These are embedded in SQP through its merit function and composite cost function, in which general nonlinear functions including unilateral, time-dependent, and nonholonomic, can be incorporated in a unified approach. Also, the avoidance of the discontinuity problem with unilateral constraints is due to the time-dependent constraints strategy, which actively adapts to varying states. Numerical examples using multibody dynamic models of a redundant manipulator demonstrate these advantages.

KW - Constraint prioritization

KW - Controlled infeasibility

KW - Normalization

KW - Optimal motion planning

KW - Redundant robot

KW - Sequential quadratic programming

UR - http://www.scopus.com/inward/record.url?scp=84874909735&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84874909735&partnerID=8YFLogxK

U2 - 10.1016/j.mechmachtheory.2013.02.001

DO - 10.1016/j.mechmachtheory.2013.02.001

M3 - Article

VL - 64

SP - 155

EP - 174

JO - Mechanism and Machine Theory

JF - Mechanism and Machine Theory

SN - 0374-1052

ER -