search

hungpham2511 / toppra

robotic motion planning library
https://hungpham2511.github.io/toppra/index.html
MIT License
603 stars 167 forks source link

Question on Spline Parametrizer #228

Open leonardoedgar opened 1 year ago

leonardoedgar commented 1 year ago

Hi, on this line, may I know why the magic number 5 is used? https://github.com/hungpham2511/toppra/blob/f8c55e376e40e3f2d5eac0b75d6b82603fe74be6/toppra/parametrizer.py#L181

I understand that delta_t will be set to that value when sdot is very small but is there any specific reason why 5 is chosen? Could the magic number be 2, 3, or 4 as well? If there are no implications by selecting lower number, it might improve the final trajectory duration (e.g. selecting 4 can reduce the final trajectory duration by 1 sec)

hungpham2511 commented 1 year ago

Hi you shouldn't hit this line in most cases. When you do there is most likely an issue.

Can you provide more information on when do you have this?

leonardoedgar commented 1 year ago

Hi Hung, actually I had this issue quite frequently when I retimed the final trajectory from ParabolicSmoother. I replicated your pick-and-place scenario in one of your papers. The difference is the suction cup model and the objects' dimensions and weights.

hungpham2511 commented 1 year ago

I retimed the final trajectory from ParabolicSmoother

Ah yes, I did have some issue with the ParabolicSmoother before.

Do you have a reproducible example?