A and B are changing

[Tianfneg]

Thanks for the answer, I would like to understand that if the A and B matrices are varying with path curvature and velocity, can Pinf still be solved only once?

[plancherb1]

Unfortunately, if you have a varying A and B, you'd need to solve for a series of Ks and Ps. That said, if you have enough memory you could store a batch of them to use online.

[Tianfneg]

Thanks a lot for your reply, I think it might be advantageous to be able to buffer some of the data based on interval data

[Tianfneg]

我们也很乐意在我们的代码库中支持这种更通用的情况，所以如果您碰巧实现了它，请提交 PR，我们也非常乐意支持该选项！
Could you please share the UAV state parameter model you used? Because I'm a bit confused as to whether the parameter model for the drone doesn't vary with trajectory or speed?

[xkhainguyen]

我们也很乐意在我们的代码库中支持这种更通用的情况，所以如果您碰巧实现了它，请提交 PR，我们也非常乐意支持该选项！
Could you please share the UAV state parameter model you used? Because I'm a bit confused as to whether the parameter model for the drone doesn't vary with trajectory or speed?

Please check https://github.com/TinyMPC/tinympc-julia/blob/main/examples/interactive_quadrotor_ext.ipynb. We provide a detailed example of MPC on a quadrotor (real Crazyflie model that we've been using). Basically we linearize the nonlinear dynamics around the hovering point to obtain a LTI model with A and B matrices which remain unchanged for the entire program.

[Tianfneg]

Hello, can you give a short answer about the Xref_data data being in absolute coordinate system? I input the data in absolute coordinate system and change A,B to my vehicle model, but when tracking the y-direction the number is getting smaller and smaller almost to 0.

[xkhainguyen]

Could you describe your problem in detail in another post?

[Tianfneg]

Hello, the top three figures in each grid are the desired trajectories I was given, and the bottom three figures are the output from our tracking process. Is it because I entered the wrong type of reference trajectory?