Feature/mimic traj #361
Feature/mimic traj #361
Conversation
Conflicts: src/prpy/planning/cbirrt.py
…lization. Added logic to get mimic trajectories. Changed parameter ordering to ensure parameters read in proper order
Conflicts: src/prpy/planning/cbirrt.py
There was a problem hiding this comment.
I think we should add an argument for list of mimic_bodies instead of save_mimic_trajectories, and do the same process when the list is nonempty. That way, we don't have to loop over all bodies in traj_bodies, and checking for number of dofs (in line 246) would be easy since we know what the goal_config would contain.
| @@ -163,7 +164,7 @@ def PlanToTSR(self, robot, tsr_chains, smoothingitrs=100, **kw_args): | |||
|
|
|||
| def Plan(self, robot, smoothingitrs=None, timelimit=None, allowlimadj=0, | |||
| jointstarts=None, jointgoals=None, psample=None, tsr_chains=None, | |||
| extra_args=None, **kw_args): | |||
| extra_args=None, save_mimic_trajectories=False, **kw_args): | |||
| """ | |||
| @param allowlimadj If True, adjust the joint limits to include | |||
There was a problem hiding this comment.
Add docstring for save_mimic_trajectories
There was a problem hiding this comment.
Couldn't this also take a dict as an argument and just save the trajectory in there, i.e. a user-supplied out-arg? Then the user could do what they want with the trajectory.
| @@ -287,6 +298,8 @@ def Plan(self, robot, smoothingitrs=None, timelimit=None, allowlimadj=0, | |||
| contextlib.nested(*mimicbody_savers), save_dof_limits(robot): | |||
| response = problem.SendCommand(args_str, True) | |||
|
|
|||
| with CollisionOptionsStateSaver(env.GetCollisionChecker(), options): | |||
| response = problem.SendCommand(args_str, True) | |||
There was a problem hiding this comment.
This block is redundant. See line 299.
| if len(jointgoals) > 1: | ||
| is_endpoint_deterministic = False | ||
|
|
||
| raw_input("Continue?") |
There was a problem hiding this comment.
This needs to be removed.
Typically we plan with CBiRRT and we are only concerned with the trajectory of the robot. However, we can perform constrained planning where the robot and an object move together. Our example case has been HERB opening the frig. Therefore this extracts the non-robot trajectories from CBIRRT. To not disrupt the prpy planning pipeline, the class saves the other trajectories such that the user can grab them later.
There is an unresolved issue that I don't know how to deal with, detailed on line 239.