[Ipopt] use of IPOPT for humanoid robots

[Andreas Waechter]

Hi Sebastian,

Please send emails regarding Ipopt to the maling list (I'm forwarding my 
reply to there).  Unfortunately, my time for Ipopt support right now is 
rather limited, and others might be able to answer sooner than me.

In the Ipopt implementation it is currently not possible to add 
constraints during the solution process.  And Ipopt the only constraints 
Ipopt is guaranteed to always satisfy are bound constraints (but note the 
bound_relax option!).  Sometimes, what people do to model constraints 
where values cannot be evaluated for certain numbers is to add auxilliary 
constraints, e.g., if you have sqrt(x-y) >= z, you can write

sqrt (s) >= z
s = x-y
s >= 0 (or >= 1e-10)

Not sure if that helps....

Regards,

Andreas

PS: I'm haooy to hear you have been using Ipopt for 4 year modeling 
humanoid robots.  If you have a few minutes spare, it would be great if 
you could add a paragraph on the Success Stories wiki page for Ipopt:

https://projects.coin-or.org/Ipopt/wiki/SuccessStories

Thanks :-)

On Tue, 13 Apr 2010, sebastien lengagne wrote:

> Dear Andreas Wächter,
>
> I am Sébastien Lengagne, I am working on the multi-contact motion generation 
> for humanoid robots as a post-doctoral at the JRL (Joint Robotics 
> Laboratory), at Tsukuba (Japan) and I use IPOPT to solve motion planning 
> problem for humanoid robot for 4 years.
>
> I have some troubles with the convergence of the algorithm due to the 
> specificity of my problem. To plan a motion, I implement constraint on the 
> body position and also on the contact forces. But the contact forces have a 
> physical meaning only when the body are at the contact location.
>
> So, I would like to know if it is possible to not take into account some 
> constraints ( on the contact forces) untill some other constraints (on the 
> body position) are satisfied ?
>
> thank you
>
> -- 
> Sébastien LENGAGNE, Ph.D
> CNRS-AIST JRL(Joint Robotics Laboratory), UMI3218/CRT
> INTELLIGENT SYSTEMS RESEARCH INSTITUTE (IS), AIST
> AIST Central 2, 1-1-1 Umezono, Tsukuba, 305-8568 JAPAN
> tel : +81-29-8626508
> fax : +81-29-8626507
> mail : sebastien.lengagne at aist.go.jp
>
>
>