Closed wxmerkt closed 4 years ago
In GitLab by @nmansard on Feb 19, 2019, 09:36
this should be in unittest, even if it is an integrative test. You will then create a similar example in example for locomotion.
In GitLab by @nmansard on Feb 19, 2019, 09:37
I can merge but would prefer that you move your file to unittest with a true assert to check some few aspects of the generated movement before.
In GitLab by @cmastalli on Feb 19, 2019, 10:26
sounds a good idea. What about if I add only the CoM task for unit-testing? We could add it inside test_integrative file, or do you prefer to put in one of the defined files for unit-testing?
In GitLab by @cmastalli on Feb 19, 2019, 13:29
added 2 commits
In GitLab by @cmastalli on Feb 19, 2019, 13:59
added 37 commits
devel
In GitLab by @cmastalli on Feb 19, 2019, 14:20
added 1 commit
In GitLab by @cmastalli on Feb 19, 2019, 14:21
I created a unit-test based on the CoM task. The code is in test_quadruped.
In GitLab by @cmastalli on Feb 19, 2019, 14:21
resolved all discussions
In GitLab by @cmastalli on Feb 19, 2019, 14:25
@nmansard you can merge it now.
There are few extra work. I modified the walking problem, we can now optimize more steps. I noticed that we're not properly setting the defaultState, so the quasiStatic warm point was not working as expected. Now you have both solved this issue and added the quasiStatic warm point in the example.
Extra comment
The quasiStati warm point works well for the walking problem, however it doesn't produce any improvement for single CoM task (inside unit-test). I have no explanation of this behaviour.
In GitLab by @nmansard on Feb 20, 2019, 09:27
marked as a Work In Progress
In GitLab by @nmansard on Feb 20, 2019, 09:27
changed the description
In GitLab by @nmansard on Feb 20, 2019, 09:27
assigned to @cmastalli
In GitLab by @cmastalli on Feb 20, 2019, 09:56
@nmansard is there any reason why you didn't merge it and change it as WIP?
In GitLab by @cmastalli on Feb 20, 2019, 10:24
unmarked as a Work In Progress
In GitLab by @cmastalli on Feb 20, 2019, 17:08
added 19 commits
devel
In GitLab by @cmastalli on Feb 20, 2019, 17:09
@nmansard you can merge it now!
In GitLab by @cmastalli on Feb 20, 2019, 17:47
added 1 commit
In GitLab by @nmansard on Feb 20, 2019, 17:55
ok thx
In GitLab by @cmastalli on Feb 21, 2019, 08:19
added 1 commit
In GitLab by @nmansard on Feb 21, 2019, 12:56
mentioned in commit 23e1d65cd10c122a80e3930ee13c2282264ff2f3
In GitLab by @nmansard on Feb 21, 2019, 12:56
merged
In GitLab by @cmastalli on Feb 18, 2019, 15:30
Merges quadruped-example -> devel
In this PR I included an example with the HyQ robot (quadruped robot). This example consist of three DDP problems solved in sequences: 1) RH-RF walking phase, 2) LH-LF walking phase, and 3) CoM moving forward and backward.
An important remark here is that it's not possible to solve the entire problems with only DDP problem. Our DDP solver struggles the limitations of single shooting formulations as described here #114. Even with this 3 DDP problems, we cannot increase the number of nodes (i.e. knots) for each phase. Indeed the problems become easily infeasible to solve.
Additionally to this example, I did the follows: