Setup

git clone https://github.com/RobotLocomotion/pdc-ros.git
cd pdc-ros
git submodule update --init --recursive
cd docker
./docker_build.py

Run a simple node to stream descriptor images

(To configure, edit config/trained_networks.yaml, and then NETWORK_NAME and RGB_TOPIC at the top of pdc_ros/src/pdc_ros/pdc_ros_simple_streaming.py)

./docker_run.py
use_all
cd ..
cd catkin_ws
catkin_make
use_all
python src/pdc_ros/nodes/simple_descriptor_streaming.py 

Run a node to respond to ROS actions to find best descriptor matches

./docker_run.py
use_all
cd ..
cd catkin_ws
catkin_make
use_all
python src/pdc_ros/nodes/find_best_match_node.py 

You can edit the file pdc_ros_wrapper.py in pdc-ros and choose which demo to run (ear, tail, shoe).

1. Launch spartan and startup robot as usual.

2. In pdc-ros container launch node to respond to ROS actions to find best descriptor matches.

cd ..
cd catkin_ws
catkin_make
use_all
python src/pdc_ros/nodes/find_best_match_node.py 

3. Launch spartan_grasp container from manuelli user. Inside the spartan_grasp container do

use_ros
use_spartan_grasp
use_spartan_grasp_ros
roslaunch spartan_grasp_wrapper spartan_grasp_wrapper.launch

4. In director use graspSupervisor.test_find_best_match_and_grasp_and_stow() to execute the pick.

1. Launch spartan and startup robot as usual.

2. In pdc-ros container launch node to respond to ROS actions to find best descriptor matches.

cd ..
cd catkin_ws
catkin_make
use_all
python src/pdc_ros/nodes/poser_server_node.py 

3. Launch spartan_grasp container from manuelli user. Inside the spartan_grasp container do

use_ros
use_spartan_grasp
use_spartan_grasp_ros
roslaunch spartan_grasp_wrapper spartan_grasp_wrapper.launch

4. In director use graspSupervisor.test_run_poser(). This will collect and RGBD image at the current location, then call poser.

5. To visualize the results use graspSupervisor.visualize_poser_result()

6. Re-orient the object to the desired configuration by running grasp_supervisor.test_run_manipulate_object().