A nonlinear stiffness safe joint mechanism design for human robot interaction

Jung Jun Park, Jae Bok Song

Research output: Contribution to journalArticlepeer-review

26 Citations (Scopus)


Service robots used in human environments must be designed to avoid collisions with humans. A safe robot arm can be designed using active or passive compliance methods. A passive compliance system composed of purely mechanical elements often provides faster and more reliable responses for dynamic collision than an active one involving sensors and actuators. Because positioning accuracy and collision safety are equally important, a robot arm should have very low stiffness when subjected to a collision force that could cause human injury but should otherwise maintain very high stiffness. A novel safe joint mechanism (SJM) consisting of linear springs and a double-slider mechanism is proposed to address these requirements. The SJM has variable stiffness that can be achieved with only passive mechanical elements. Analyses and experiments on static and dynamic collisions show high stiffness against an external torque less than a predetermined threshold value and an abrupt drop in stiffness when the external torque exceeds this threshold. The SJM enables the robotic manipulator to guarantee positioning accuracy and collision safety and it is simple to install between an actuator and a robot link without a significant change in the robot's design.

Original languageEnglish
Pages (from-to)610051-610058
Number of pages8
JournalJournal of Mechanical Design, Transactions of the ASME
Issue number6
Publication statusPublished - 2010 Jun


  • Nonlinear stiffness
  • Passive compliance
  • Physical human-robot interaction
  • Safety mechanism

ASJC Scopus subject areas

  • Mechanics of Materials
  • Mechanical Engineering
  • Computer Science Applications
  • Computer Graphics and Computer-Aided Design


Dive into the research topics of 'A nonlinear stiffness safe joint mechanism design for human robot interaction'. Together they form a unique fingerprint.

Cite this