EmaMaker
eee95b0059
[UNSUCCESS] try to minimize vcorr^2
2024-09-16 18:27:41 +02:00
EmaMaker
41f0d66851
correction is on untrasformed inputs, but cost and constraints are on robot inputs
2024-09-16 18:04:18 +02:00
EmaMaker
ceb7659bcc
misc tests
2024-09-13 20:26:01 +02:00
EmaMaker
854247dee1
[IMPROVE] build A_deq outside for loop
2024-09-11 22:20:31 +02:00
EmaMaker
a9bf86cda8
big refactor of branches into a single codebase
...
using flags to switch between tests and robot types
2024-09-11 19:34:41 +02:00
EmaMaker
f90e3bdb68
control_act: make correction refer to w_r, w_l
...
This fixes a substantial bug with 1-step and multistep behaving
differently than track-only
for more information: see notes 10/09/24, but basically what was
happening was that just v^{corr} was being minimized, for how the
constraints were built
2024-09-10 21:09:32 +02:00
EmaMaker
3ddb3ec4ca
control_act: 1-step into its own if-statement
2024-09-10 21:07:32 +02:00
EmaMaker
3b0257adf2
control_act: use blkdiag to build A_deq
2024-08-28 16:12:22 +02:00
EmaMaker
6248525cbe
a nice battery of tests
2024-08-28 16:11:49 +02:00
EmaMaker
ef8a12166a
save history of predicted trajectory for video
2024-08-01 12:23:49 +02:00
EmaMaker
c02aa51088
extend unicycle to differential drive
2024-08-01 12:23:49 +02:00
EmaMaker
67167598d9
control_act: fix multi-step mpc
2024-08-01 12:23:43 +02:00
EmaMaker
1900526b81
bug: fix u_corr_history not being saved between different calls to u_corr()
...
+ plot u_track and u_corr
2024-07-26 20:13:43 +02:00
EmaMaker
89cb27ff49
restructure code to allow execution in parallel
...
these commits are so messy
2024-07-24 14:57:19 +02:00
EmaMaker
ec8b2dcecb
multi-step MPC
...
seems to be working?
2024-07-24 11:49:34 +02:00
EmaMaker
02fdac42e3
rename state x to q
2024-07-23 18:07:50 +02:00
EmaMaker
44f65aed77
1-step mpc
2024-07-16 10:58:00 +02:00
EmaMaker
c79a8744b2
simulate system in discrete time
2024-07-14 15:16:05 +02:00
EmaMaker
6d84703c62
isolate control in its own function
2024-07-13 11:09:40 +02:00
EmaMaker