Notes from Dan's Presentation at PI Conf. Call April 1 2004

Thanks for talking with people at PI telecon. It was clear that people were interested and it was good to hear you talk about it.

Please ask Will for his copy of the World Haptics proceedings. There was an interesting short paper or poster session about how to do "shading" between the facets of a friction cone or pyramid. I hope it's in the proceedings in some form...

Regarding Dan K's question: I have not seen the "embedded cone" in the literature, at least not exactly as we've described it. We should investigate this. And also look at other friction models & see how our current thinking agrees with it. We may have something a bit novel. Or maybe not.

Remember also our thoughts about Tresca or Von Mises yield criterion and whether that would provide a nice explanation for the angle of the embedded part of the cone. Idea: combined (tensile + shear) loading => angle of embedded part of cone. (Look in a good strength of materials or basic plasticity like Johnson & Mellor or others. Will can help you find a good ref.). Tresca will be linear... Von Mises will be more ellipsoidal...

Other nice extensions are that the cone can be non-symmetric (anisotropic friction) and there was also our thought about modeling the orientation that a spine contacts an asperity. Assume Hertzian type contact... but make the orientation be the local surface normal orientation on the asperity bump.

Cool stuff for possible ICRA paper.

>>> Sounds like we need to see Al Rizzi's paper!!! Please get it from him.

Please don't oversell the linear programming problem. The Kerr and Roth approach is no longer state of the art. Others have since formulated more or less the same problem in ways that are faster & better.

There's also a new Ph.D. thesis from Constanineau (sp?), a student of Tim Salcudean at U. British Columbia, who has done a very comprehensive literature review of recent friction models and contact models. We (you) need to get this. I was an external reviewer at her defense and was impressed with her lit. review.

In general, the Haptics community is, I think, ahead of the general dynamics/simulation community in terms of modeling friction appropriately. That's because it is so important for haptics. Brent Gillespie may also have some recent stuff.

Personally I like the hybrid discrete/continuous system approach. My conversations with Brent have continued to make me in favor of it. Still there are many tricky issues. We also need to touch base with Khatib's students & see what they are up to more recently.

Good points from Al about difficulties of getting any kind of force control from RiSE. Compliance will help. Ankle instrumentation will be essential. That way, even if we can't ideally control the forces we can at least measure what we actually got in terms of forces and see if that makes good sense in terms of the analysis.

I am putting these notes on TWiki page so we can all see them during group debriefing after conf. call. Other more basic Arachi simulation/graphics/integration issues are a bit beyond the scope for us, at least in the short term...

http://bdml.stanford.edu/twiki/bin/view/Main/NotesFromSantosTalkApril2004

-- MarkCutkosky - 01 Apr 2005