Human-Robot Cooperative Manipulation Using a Virtual Nonholonomic Constraint

Abstract

A robotic assistance system is presented which enables a human to handle unwieldy long objects by providing only one point of support. The robot grasps one end of the object and helps the human operator carry it at the other end. The proposed control method uses a virtual nonholonomic constraint. The movement of the object is constrained as if it were being carried on a wheel attached to the object. This method prevents the object from slipping sideways and simplifies the carrying operation. In spite of the constraint, the object can reach any position and orientation due to the nonholonomy. Cooperative manipulation in a horizontal plane is considered first, and then it is extended to 6-DoF manipulation in 3-D space. Experimental results show that an operator can easily handle a long object when aided by the robot.

The role of roles: Physical cooperation between humans and robots

“…This approach focuses on desired paths or workspace constraints rather than desired virtual dynamic object behavior, similar to the virtual fixtures introduced by Rosenberg (1993) as overlays such as virtual rulers guiding the operator's effector motion in telepresence setups. An approach combining desired virtual constraints and desired virtual object dynamics was proposed by Takubo et al (2002). In their work, a robotic partner rendered a virtual non-holonomic constraint, namely a virtual wheel, that prohibits sideway slipping motion and thus simplifies operation, similar to a wheelbarrow.…”

Human-Robot Cooperation in Precise Positioning of a Flat Object

“…In Yokoyama et al [September 14-19, 2003] the modes are switched by human command. In Takubo et al [May-June 2002] the problem was tackled indirectly. By using nonholonomic constraints the object Seoul, Korea, July 6-11, 2008 Fig.…”

Replicating Human-Human Physical Interaction

“…Griffin et al [11], Griffiths and Gillespie [11], and Takubo et al [12] have demonstrated robots physically working with humans, but their controllers are not based upon the interactions of two people. Their methods predominantly try to stabilize the system while the human manipulates the object.…”