Integrate toppra for time parameterization.

[Levi-Armstrong]

https://github.com/hungpham2511/toppra

[mpowelson]

@marrts might have some useful code around somewhere. This was discussed in this crs pull request.

[mpowelson]

Also, I just noticed that though the license is MIT, they import ECOS which is GPL. Doesn't that mean that it would have to be GPL as well?

Edit, I think ECOS is only in the python code. The new C++ API looks to use qpOASES

[gavanderhoorn]

Edit, I think ECOS is only in the python code. The new C++ API looks to use qpOASES

Exactly. I was just going to post a comment about this.

And it's all configurable as well.

[marrts]

@marrts might have some useful code around somewhere. This was discussed in this crs pull request.

I didn't look too much into toppra at that time and I ended up using this modified code from moveit's iterative spline parameterization from @schornakj which has served well for me but doesn't control for tool speed at all, just joint velocity and acceleration.

[Levi-Armstrong]

If they are integrating it using qpOASES could we just introduce it as a cost/constraint to our trajopt problem with the new IFOPT integration work? This way it may influence the final trajectory instead of a post trajectory operation.

[mpowelson]

I think I know the answer to this, but has anyone by chance looked into this? TOPPRA is getting some votes of confidence over on moveit/moveit#2183.

[Levi-Armstrong]

I had looked at it a few months back, but have not testing anything.