Shuo

1. in convex MPC, an astonishing fact is - the actual value of inertia does not matter because when you tune the Q and R matrices, they implicitly account for...

(sorry this computer I am using has no pinyin input methods) 1. I observed the same problem. The simulation of the main branch is not fine tuned. I will check...

Let me know how it works.

2. I fixed this bug. the use_sim_time is not correctly loaded and checked so the hardware code uses the wrong control dt. Now the performance on hardware should be good.

I will tune the MPC gazebo gain a little bit. MPC should always performs better than QP when fine tuned.

The same way as A1

Thank you! Feel free to submit a PR

OK, I did not read the second post carefully. Thanks!

You have to make a new config folder as `https://github.com/ShuoYangRobotics/Cerberus/tree/main/config` And put your camera parameters and robot transformation parameters in it. Moreover, the leg kinematics in Cerberus is hand coded....

Yes, position Z is not fine-tuned and hard to be estimated correctly. Since we have other height estimation method (barometer), we do not use Cerberus for Z height estimation