Skip to content

Latest commit

 

History

History
72 lines (52 loc) · 4.24 KB

README.md

File metadata and controls

72 lines (52 loc) · 4.24 KB

Standard reach study

This exercise will show you how to run and re-configure a "standard" reach study, which involves a robot reaching target poses on a workpiece

Problem statement

Evaluate the reachability of a Motoman SIA20D for inspection of a vehicle hood panel

Tasks

1. Run the demo in the reach_ros package

  • Run the "setup" launch file, which loads the URDF and starts the ancillary ROS nodes (Rviz, joint state publisher, etc.)

    ros2 launch reach_ros setup.launch.py

  • Run the reach study, loading pre-generated results

    ros2 launch reach_ros start.launch.py

  • Left-click on the interactive markers to see the robot pose at each Cartesian target

  • Right-click on the interactive markers to see the score of the robot

2. Run the reach_ros demo again using the same configuration file, but generate the results from scratch to a new destination

  • With the "setup" launch file still active, re-run the "start" launch file with new arguments

    ros2 launch reach_ros start.launch.py results_dir:=/tmp config_name:=study_2

    How many iterations did it take the study to converge?

  • Open the reach study database (/tmp/study_2/reach.db.xml) and introspect its contents

3. Generate a heat map for the results of the reach study

  • Use this script to generate a heat map for the results of a reach study

    python3 heat_map_generator.py /tmp/study_2/reach.db.xml `ros2 pkg prefix reach_ros`/share/reach_ros/demo/config/part.ply

    Note: this script can be found in reach repository, however there is small bug with the current version. See this issue.

    Note: the generalized command for this script is python3 heat_map_generator.py <reach database file> <workpiece mesh file>

    Note: the mesh provided to this script should be the same mesh from which the reach study target poses were generated

4. Relax the constraint on the orientation of the tool z-axis and re-run the reach study

In the case that our reach study only requires that the tool z-axis to align with the surface normal (e.g., for processes like sanding, painting, inspection, etc.), we would like to relax that constraint in the reach study and see how it affects the results. We have a few options for doing this:

  1. (recommended) Use the provided DiscretizedMoveItIKSolverFactory as the REACH IK plugin. This involves modifying the reach study configuration YAML file.

    Note: The DiscretizedMoveItIKSolverFactory plugin has been renamed to DiscretizedMoveItIKSolver after version 1.4.0 of reach_ros2

  2. Use a different MoveIK solver plugin that allows orientation constraints to be set. An example is this modification to the trac_ik MoveIt plugin. This change involves a modification to the MoveIt kinematics.yaml file, like the one here.
  3. Create a custom implementation of the TargetPoseGenerator interface that creates multiple orientations (each rotated about the surface normal) for each target position. This is essentially the same as option 1, but requires that we create a new class.

5. Change the evaluation metrics of the reach study

Maybe maximizing "raw" manipulability and distance from nearest collision are not objectives that we care about. Instead, change the objectives such that the reach study:

  • Maximizes the ratio of manipulability for position only
    • This effectively means that we want the robot to be able to translate in all directions equally effectively
  • Tries to keep the joints as centered in their joint range as possible

This changes involves modifying the reach study configuration YAML file. Check out the description of the existing evaluator plugins to figure out which plugins accomplish the objectives defined above. Re-run the reach study, saving the results to a new file, and see how the results differ from the first reach study