Alexis DUBURCQ
Alexis DUBURCQ
I'm going for a full refactoring a the log reading/writing procedure (both C++ and Python) to make it easier to manipulate it, avoid edge-cases raising exceptions, and reduce boilerplate.
In principale, i should make jiminy/pinocchio code jit-able with JAX as a side effect, which is a tremendous advantage.
After looking more closely at the [implementation of the cholesky decomposition](https://github.com/stack-of-tasks/pinocchio/blob/46abd97c9738e05b7ffafa1b0e0a7011f12cf2ca/src/algorithm/contact-cholesky.hxx#L266) in Pinocchio, it is quite far from the current architecture of Jiminy. Jiminy is providing the [whole jacobian as...
Jiminy has been refactored as planned. Now it would be possible to implement the Cholesky decomposition of `JMinvJt` as proposed in the [paper](http://www.roboticsproceedings.org/rss17/p017.pdf).
Yes me too, but I'm afraid as @rdb said, it is impossible without the new shader compilation pipeline in 1.11, so it is going to take a long time as...
Easy, just never call `set_shader_auto` nor `set_shader_off` anywhere at all ! But obviously the rendering will be much less beautiful.
Yes, I warned you about that. I did not find any better option so far unfortunately.
I can check but I'm pretty sure it does not work. I tested in on 21.10, not 22.04. Regarding the first issue, the point is modern intel GPU does not...
For the second bug, adding **multiple** `sim.graphics_engine.render_frame()` before `images = sim.get_images()` fixes the issue. I also noticed that after adding tons of `render_frame` calls (like 15) it takes a while...
Apparently I managed to fix all the issues by calling `render_frame` right after adding a offscreen camera systematically, and once before calling `get_images`. I will check on another machine to...