-
Notifications
You must be signed in to change notification settings - Fork 569
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix: ensure attached objects update during motion execution #3327
base: main
Are you sure you want to change the base?
fix: ensure attached objects update during motion execution #3327
Conversation
- Check that attached objects in the monitored robot match those in the planned trajectory. - If an object disappears from the monitored robot, remove it from the trajectory waypoint robot_state. - If an object is attached to the monitored robot but missing in the trajectory, add it to enable meaningful collision checking.
auto getAttachedObjects = [](const moveit::core::RobotState& state) { | ||
std::vector<const moveit::core::AttachedBody*> attached_bodies; | ||
state.getAttachedBodies(attached_bodies); | ||
std::map<std::string, const moveit::core::AttachedBody*> attached_objects; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Does it make sense to add this function to RobotState
itself instead? Like a getAttachedBodiedMap()
function?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That sounds like a good idea! Adding a getAttachedBodiesMap()
function to RobotState could make it more reusable and improve clarity. Would you like to propose this as a separate PR, or do you plan to integrate it into your current changes?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I was suggesting including this in your PR, if you would be up for it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have just created the getAttachedBodies
version that returns a map instead of a vector and updated the PR accordingly. Let me know if this aligns with your ideas!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It looks good, though I would consider not having it call the other one since it seems inefficient to create a vector, then loop through that to create a map. Maybe just go straight by looping through and putting together the map?
for (std::size_t i = std::max(path_segment.second - 1, 0); i < wpc; ++i) | ||
{ | ||
state = t.getWayPoint(i); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you for this!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The only simplification I could think of is that, under the assumption that "attached objects are set to the waypoint's robot state at planning time" (which actually holds) we could query sample_attached_object only once. However, I proposed updating it for each waypoint to ensure robustness, even if this assumption doesn't hold. Do you see any potential issues or improvements with this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah thanks for asking.
I guess I would be thinking about:
- Does it slow things down to be potentially attaching/detaching objects at each waypoint during this check?
- It's difficult for me to know what the user's intent would be on a case by case basis. Right now it seems like the current state is treated as the source of truth, but I'm unsure whether users want that in every case. I guess the previous implementation did the opposite and treated the attached objects from the pre-planned waypoint as the source of truth.
Is it worth maybe adding a flag to this function for whether one wants to prioritize current state vs. planned states? And elevate this up to the config/parameter level?
@@ -283,34 +283,60 @@ bool plan_execution::PlanExecution::isRemainingPathValid(const ExecutableMotionP | |||
collision_detection::CollisionRequest req; | |||
req.group_name = t.getGroupName(); | |||
req.pad_environment_collisions = false; | |||
moveit::core::RobotState state = plan.planning_scene->getCurrentState(); | |||
std::map<std::string, const moveit::core::AttachedBody*> current_attached_objects, sample_attached_object; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nit: sample_attached_objects
(with an s at the end) for consistency.
Codecov ReportAttention: Patch coverage is
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## main #3327 +/- ##
==========================================
- Coverage 45.59% 45.56% -0.03%
==========================================
Files 716 716
Lines 62400 62465 +65
Branches 7548 7563 +15
==========================================
+ Hits 28445 28454 +9
- Misses 33788 33844 +56
Partials 167 167 ☔ View full report in Codecov by Sentry. |
Description
Implications
This could impact a few different situations: