Newsgroups: comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!howland.reston.ans.net!Germany.EU.net!EU.net!uknet!festival!hwcee!news-serv!acc
From: acc@dibble.cee.hw.ac.uk (Andrew Clegg)
Subject: [RESEARCH Q] How do *you* model manipulator joint friction ?
Message-ID: <ACC.95Mar2171516@dibble.cee.hw.ac.uk>
Followup-To: comp.robotics
Sender: news@cee.hw.ac.uk (News Administrator)
Reply-To: acc@cee.hw.ac.uk
Organization: Ocean Systems Lab. Heriot-Watt University, Edinburgh
Date: Thu, 2 Mar 1995 17:15:16 GMT
Lines: 48

[ Just a try to stimulate some research discussion ]

How do you model the joint friction of a manipulator ? And what
parameter values have you found to be realistic ? 

I use a simple model with viscous and Coulomb friction components
as :- 

	joint_friction = V theta_dot + C sign(theta_dot)

where theta_dot = joint angular velocity (rotational joints)
      V = viscous friction coefficient
      C = Coulomb friction coefficient
      sign = Signum function (I'd sooner use sgn - but sign is the
	     Matlab convention)

However, when I include a non-zero value for C the simulation grinds
almost to a halt. [ My models/simulations are run under Matlab with
Simulink, using the Runge Kutta Fifth order integration routine. ]

The method that Peter Corke uses in the Matlab Robotics Toolbox is :-

	joint_friction = V theta_dot + CP (theta_dot>0) +
				CN (theta_dot<0)

This allows the Coulomb friction to be asymmetric, CP for positive
velocities and CN for negative velocities. This model takes longer to
simulate than the first one.

Any body use any other forms ? How about Stiction - where the static
Coulomb friction is greater than the value when it is moving, e.g. :-

	if theta_dot==0			% or some range close to zero ?
		joint_friction = CS
	else
		joint_friction = V theta_dot + C sign(theta_dot)
		
where CS is the Static Coulomb friction coefficient, and CS > C.

Just a few thoughts. Any comments anybody ?


Andy Clegg________________________________________acc@cee.hw.ac.uk___

Ocean Systems Lab, Dept of Computing & Electrical Engineering,
Heriot-Watt University, Edinburgh, EH14 4AS  Tel: (+44) 0131 451 3506
http://www.cee.hw.ac.uk/~acc/acc.html	     Fax: (+44) 0131 451 3327
_____________________________________________________________________
