Tobia Marcucci
Tobia Marcucci
I think that either throwing an error or a warning when a self-edge is added is good. On a similar note, I think that it'd be helpful to have an...
I just spoke with Russ, and we agreed that self-edges can never be optimal. However, we also think that rather than removing self-edges from the optimization problem, it'd be more...
@bernhardpg The reason why it is not optimal is exactly what you write in the second part of you message: because the continuous variable is only one.
@RussTedrake Thinking more about this though, I think that the convex relaxation can get looser by adding self-edges. I've an example in mind similar to the one in the paper....
I would disallow it at the GCS level and, yes, it should be carefully explained why it is disallowed (it could be handled fine, but it weakens the formulation and...
Yes, let's just double check that the relaxation can actually get weaker when there are self edges.
I see. For the second point, they've always sent us screenshots or a copy and paste of the error message. So I think it's safe to just delete the json....
I would not say that Lorentz Cone is a subset of Rotated Lorentz Cone. They are the same thing except for a rotation matrix. I think by doing what you...
@iamsavva i don't think that's correct. SOCs and rotated SOCs are the same families of sets, but one has an invertible rotation matrix in front. They have the same representation...
Yes, my goal as said is to linearize the dynamics of a `RigidBodyTree` around a point that is not an equilibrium, and I'm working in python. At the moment I'm...