Holonomic vehicle with orientation: regularization

[timmercy]

It is possible to regularize on the change in orientation of the vehicle. For the examples I tested this was only necessary for the freeT problems. fixedT problems solved well without regularization.

For now there is an option which the user can set to regularize with a 1-norm or a 2-norm. He has to tune the weight himself.

There are 2 issues here:

• If you do a 1-norm regularization with weight 0 there are still constraints added, so while you would expect no change in behaviour there is really a change.
• If you do a regularization for the fixedT problem with a weight which is too high you reach the goal position, but not with the desired orientation.

[rubenvanparys]

You could check if the weight == 0.0 in order to add the constraints and
part of the objective.

The second issue sounds natural as a trade-off is made between 'minimize
change of orientation' and 'bring orientation to desired orientation'...

On 13-04-16 11:13, timmercy wrote:

It is possible to regularize on the change in orientation of the
vehicle. For the examples I tested this was only necessary for the
freeT problems. fixedT problems solved well without regularization.

For now there is an option which the user can set to regularize with a
1-norm or a 2-norm. He has to tune the weight himself.

There are 2 issues here:

• If you do a 1-norm regularization with weight 0 there are still
  constraints added, so while you would expect no change in
  behaviour there is really a change.
• If you do a regularization for the fixedT problem with a weight
  which is too high you reach the goal position, but not with the
  desired orientation.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#18

[timmercy]

Okay yes, that's maybe the most safe way to implement it then, I will push it like that.