-
Error when model default bounds are set (?), seems iLQG is not checking for bounds correctly and boxQP throws an error.
@vikashplus
-
Hello everybody!
Recently I have managed to optimize the Humanoid Tracking task to run in real time (on 16 thread cpu ) by increasing physics time step size from 0.002 sec upto 0.016 sec . Howev…
-
On [this line](https://github.com/google-deepmind/mujoco_mpc/blob/v0.0.1/mjpc/planners/ilqs/planner.cc#L120), local variables `dim_actions` and `dim_parameters` are declared, which means that the assi…
-
My setup was:
```
user@Macbook-MacBook-Air-4 mujoco_mpc % clang --version
Homebrew clang version 14.0.6
Target: x86_64-apple-darwin20.6.0
Thread model: posix
InstalledDir: /usr/local/opt/llvm@14…
-
Dear Deepmind Team, after I cmake the project, I found some issues when I tried to use this lib on my Ubuntu20 system. I ask someone to do the same test and the problem are the same. So I would like t…
-
In iLQG planner, the [nominal policy is updated](https://github.com/deepmind/mujoco_mpc/blob/main/mjpc/planners/ilqg/planner.cc#L578) with `candidate_policy[winner]` where `winner` is the [index](http…
-
The iLQG nominal rollout, defined [here](https://github.com/deepmind/mujoco_mpc/blob/15b4385b6e7a662d4f37ada404aead2b16e0c3ce/mjpc/planners/ilqg/planner.cc#L344), is inefficient. We should do this wit…
-
I have tried to build _all_ ROS packages on Linux and it's going quite smoothly for a lot of them.
Some do need some "fixing" though (a patch here and there) and this is the list of those (some are…
wolfv updated
2 years ago
-
- A `FeedbackMotionSolver` only returns the control trajectory as a solution, while the gains remain transient in that instance of the solver.
- This makes loading and replaying control policies impo…
-
The `Solve` methods are too long, spanning a couple of screens. Should be refactored.