costar_stack

Integrated ROS capabilities for planning, predicate inference, gripper control, and perception for use with the KUKA LBR IIWA and Universal Robots.

View the Project on GitHub cpaxton/costar_stack

Collecting Additional Data

Goal: to collect data for the white bin in which we will be building our tower.

We run the object on table segmenter to collect this data:

roslaunch object_on_table_segmenter object_on_table_segmenter.launch object:=white_object table_tf:=ar_marker_2 load_table:=true auto_capture:=false

List of Commands

Bring these up in separate windows (I’m working on a script for this):

roscore
roslaunch costar_bringup ur5_c_model.launch use_scene_parsing_pose:=true \
  use_external_segmenter:=true
roslaunch color_nn_segmenter costar_segmenter.launch load_table:=true \
  load_existing_model:=true model_name:=rgby_wood_white_bin \
  background_labels:=wood_block,white_bin \
  foreground_labels:=blue_block,green_block,red_block,yellow_block \
  kmeans_point_per_model:=2 table_tf:=ar_marker_2
roslaunch sp_segmenter colored_block.launch
roslaunch sequential_scene_parsing block_scene.launch best_hypothesis_only:=true

Outputs

This produces a set of data in the local folder. These point clouds should include the pixel values that get grouped together as “white.”

Color Segmenter

We want to use the color segmenter for our block-stacking task.