snp_automate_2022: generate motion plan failure #40
Comments
|
Is this running on the master branch of this repo? I'd personally recommend running on my PR here as the debugging is greatly improved. If you're running on this branch already I'd recommend running it with verbose on to assist with debugging further: One useful feature is that it generates a graph of the planning pipeline with the taken path highlighted. With the verbose tag set to true it'll automatically save this file to Overview of Full Graph:
Top Level Pipeline:
Raster Taskflow Pipeline:
A Single Motion Plan Pipeline
From this I can see that the raster planning taskflow failed ultimately because the |
|
Did you run into any issues with setting up the stack on the branch you linked? I'm running into an error with I fixed the fcl build issue above by following the post. I've also built the octomap package using the build instructions on their github page and set the correct OCTOMAP_DIR. However, ros can't seem to find the package ( |
|
Have you tried installing |
|
Yup I did, I got the same error on tesseract_geometry and ros2 pkg prefix still returned empty note that this didn't happen on the master branch, it occurs on #22 |
The error you're reporting above is CMake not being able to find |
|
Okay so I'm running on the master branch and was able to run the example demo in automate_2022. I swapped out the default robot for our custom robot. After some debugging, looks like the motion plans are now failing due to IK failures. I updated the IKFast plugin configuration with the updated frame names, etc. We have the analytical solution for our robot's IK. Could someone point me to how I can create an IK plugin that allows me to use my analytical solution? Or alternatively is it better to simply add my IK solver directly into the descartes_robot_sampler's |
Thank you! How can I go about editing the relative pose of the mesh object with respect to its reference frame? (I see that the marker is publisheed in open3d_sim_node.cpp, but I'm assuming there's purely visualization and there's some other code that stores its pose) |
|
The pose is relative to the reference_frame parameter. If you want to keep the same reference frame but move the mesh you can use meshlab to modify the mesh directly. |
|
Are there any explicit formatting options for the mesh file? I have a mesh file for a table in the correct format (.ply) but the tool can't seem to generate the raster points. I tried different outlines using the polygon selection tool. The error in the rviz panel is either "no points found within polygon boundary" or "no clusters found". The mesh file looks okay when opened in meshlab I see this warning in the terminal: mobile_base_link is a valid frame in my system though (verified in the rviz tf tree) and I don't get this error when using the default part_scan.ply mesh file
|
Not particularly. I have noticed that .ply files exported from Solidworks sometimes can't be displayed (possible error in formatting from Windows to Linux). But if it shows up in Rviz, it loaded the mesh correctly.
This is the real error. One step in segmenting the region of interest submesh with the boundary is clustering the vertices inside of the boundary. There is a parameter that controls the minimum cluster size (100, by default; see image below). If there are not at least that many mesh vertices inside the selection boundary, then it won't identify the sub-mesh. I would try reducing that value and see if it generates tool paths
|
I'm running the stack on the foxy-desktop docker image
after launching snp_automate_2022 through
ros2 launch snp_automate_2022 start.launch.xml, I went through the steps in the rviz plugin:(1) and (2) pass but I get a failure on (3), logs_A attached below. I haven't changed any of the mesh/tool planning settings through the other gui that opens up.
Next, I played around with the raster planner parameters and changed the following values
the motion plan status now says CONVERGED but there's still a failure at the end (logs_B)
logs_A
logs_B
The text was updated successfully, but these errors were encountered: